بایگانی برچسب برای: dv

به خاطر دارید که Delegate نوع های داده ای بودند که اطلاعات مربوط به یک متد را در خود نگهداری می کردند؟ زمانی که یک delegate جدید تعریف می کنید، در حقیقت کلاس جدیدی ایجاد می شود که این کلاس، از کلاس MultiCastDelegate مشتق شده است. این موضوع باعث می شود که delegate تعریف شده شامل یکسری متدها باشد. قبلاً با delegate ها و شیوه فراخوانی آن ها آشنا شدیم. در این مطلب می خواهیم با شیوه فراخوانی متدها به صورت Asynchronous آشنا شویم. زمانی که از روی یک delegate شئ ای برای یک متد ایجاد می کنیم، این شئ شامل دو متد به نام های زیر است:

  1. BeginInvoke
  2. EndInvoke

از این دو متد می توان برای فراخوانی delegate ها به صورت Asynchronous استفاده کرد. اگر به خاطر داشته باشید، delegate ها یک متد دیگر نیز داشتند به نام Invoke که به وسیله این متد می توانستیم delegate را به صورت عادی فراخوانی کنیم. تفاوت Invoke و BeginInvoke در این است که متد Invoke عملیات فراخوانی را در Thread جاری انجام می دهد، اما متد BeginInvoke یک Thread جدید ایجاد کرده و delegate را در آن Thread فراخوانی می کند. برای آشنایی بیشتر با یک مثال جلو می رویم، Delegate ای به صورت زیر تعریف می کنیم:

public delegate int MathOperation(int n1, int n2);

در ادامه کد زیر را نیز اضافه می کنیم:

static void Main(string[] args)
{
    MathOperation operation = new MathOperation(Add);
    Console.WriteLine("Main thread ID:"+Thread.CurrentThread.ManagedThreadId);
    operation(2, 6);
}

public static int Add(int n1, int n2)
{
    Console.WriteLine("Add thread ID: " + Thread.CurrentThread.ManagedThreadId);
    return n1 + n2;
}

به خطوط 4 و 10 دقت کنید، در این خطوط از کلاس Thread استفاده کردیم، کلاس Thread یک Property دارد به نام CurrentThread که اطلاعات Thread جاری که متد در آن اجرا شده است را بر میگرداند، خاصیت CurrentThread از نوع کلاس Thread است که بوسیله خاصیت ManagedThreadId می توان شناسه Thread در حال اجرا را بدست آورد. بعد اجرای کد بالا با خروجی زیر مواجه می شویم:

Main thread ID:1
Add thread ID: 1

همانطور که مشاهده می کنید شناسه Thread بر هر دو متد Main و Add یکسان است، اما همانطور که گفتیم می خواهیم متد Add را در یک Thread جداگانه اجرا کنیم. در اینجا از متد BeginInvoke که برای Delegate تعریف شده استفاده می کنیم. برای MathOperation متد BeginInvoke به صورت زیر تعریف شده :

IAsyncResult BeginInvoke(int n1, int n2, AsyncCallBack callback, object state);

همانطور که مشاهده می کنید پارامترهای اول و دوم تعریف شده برای BeginInvoke مبتنی بر پارامترهایی است که برای Delegate تعریف کردیم. پارامترهای callback و state را هم بعداً بررسی می کنیم، در حال حاضر برای فراخوانی متد BeginInvoke برای این دو پارامتر مقدار null ارسال می کنیم.

همچنین متد EndInvoke نیز برای MathOperation به صورت زیر تعریف شده است:

int EndInvoke(IAsyncResult result);

همانطور که مشاهده می کنید مقدار بازگشتی EndInvoke از نوع int است که بر اساس نوع بازگشتی delegate تعریف شده مشخص می شود. همچنین پارامتر ورودی EndInvoke از نوع IAsyncResult است که در ادامه به بررسی این interface خواهیم پرداخت.

اینترفیس IAsyncResult

همانطور که مشاهده کردید، مقدار بازگشتی متد BeginInvoke از نوع IAsyncResult است. این interface در متدهای BeginInvoke و EndInvoke استفاده می شود، در متد BeginInvoke مقدار بازگشتی شئ ای از نوع IAsyncResult است و در متد EndInvoke پارامتر ورودی این متد از نوع IAsyncResult می باشد. تعریف IAsyncResult به صورت زیر است:

public interfae IAsyncResult
{
    object AsyncState { get; set; }
    WaitHandle AsyncWaitHandle { get; set; }
    bool CompletedSynchronously { get; set; }
    bool IsCompleted { get; set; }
}

در ساده ترین حالت ممکن شما نیازی به کار با اعضاء این اینترفیس ندارید، تنها کاری که باید بکنید نگهداری مقدار بازگردانده شده از متد BeginInvoke و ارسال آن به متد EndInvoke در زمان مناسب برای گرفتن خروجی است. برای آشنایی بیشتر با متدهای BeginInvoke و EndInvoke و همچنین استفاده از IAsyncResult با یک مثال ساده جلو می رویم. کدی که برای MathOperation در ابتدای این مطلب نوشتیم را به صورت زیر تغییر می دهیم:

MathOperation operation = new MathOperation(Add);
Console.WriteLine("Main thread ID:"+Thread.CurrentThread.ManagedThreadId);
var result = operation.BeginInvoke(2, 6, null, null);
Console.WriteLine("Task in Main method.");
var answer = operation.EndInvoke(result);
Console.WriteLine("2 + 6 = {0}", answer);

بعد از اجرای کد بالا، خروجی به صورت زیر خواهد بود:

Main thread ID:1
Task in Main method.
Add thread ID: 3
2 + 6 = 8

همانطور که مشاهده می کنید، شناسه Thread برای متد Add مقدار 3 می باشد، به این معنی که متد Add در حال اجرا در یک Thread جداگانه از Thread اصلی برنامه یا Main Thread است. اما یک موضوع هنوز باقی مانده، Synchronization. به متد Main دقت کنید، زمانی که delegate با متد BeginInvoke فراخوانی می شود و بعد از آن پیغام Task in Main method را در خروجی چاپ می کنیم، در حقیقت Thread جاری برای لحظاتی متوقف شده و بعد بوسیله متد EndInvoke خروجی را دریافت می کنیم. اما کاری که ما می خواهیم انجام دهیم همزمانی اجرای متد WriteLine و متد Add است، برای اینکار باید از اعضاء IAsyncResult استفاده کنیم. در این اینترفیس خصوصیتی تعریف شده با نام IsCompleted که در صورت اتمام اجرای متد مقدار true را بر میگرداند. کدی که نوشتیم را به صورت زیر تغییر می دهیم:

static void Main(string[] args)
{
    MathOperation operation = new MathOperation(Add);
    Console.WriteLine("Main thread ID:"+Thread.CurrentThread.ManagedThreadId);
    var result = operation.BeginInvoke(2, 6, null, null);
    while (!result.IsCompleted)
    {
        Console.WriteLine("Task in Main method...");
        Thread.Sleep(1000);
    }
    Console.WriteLine("Task in Main method.");
    var answer = operation.EndInvoke(result);
    Console.WriteLine("2 + 6 = {0}", answer);
}

public static int Add(int n1, int n2)
{
    Console.WriteLine("Add thread ID: " + Thread.CurrentThread.ManagedThreadId);
    Thread.Sleep(5000);
    return n1 + n2;
}

به متد Thread.Sleep دقت کنید، این متد روند اجرای Thread را برای مدت زمان مشخص شده متوقف می کند. عدد وارد شده به میلی ثانیه است. در کد بالا متد Add به میزان 5 ثانیه و متد با هر بار تکرار حلقه while متد متد Main به اندازه 1 ثانیه متوقف می شوند. اجرای کد بالا خروجی زیر را تولید می کند:

Main thread ID:1
Task in Main method...
Add thread ID: 3
Task in Main method...
Task in Main method...
Task in Main method...
Task in Main method...
2 + 6 = 8

با اینکه متد Add فراخوانی شده و Thread مرتبط با آن به مدت 5 ثانیه در حالت Sleep قرار گرفته، اما متد Main در حال انجام کار خودش است و تا زمانی که متد Add کامل نشده و IsCompleted در IAsyncResult مقدار true بر نگرداند دستور WriteLine فراخوانی شده و پیام Task in Main method بر روی صفحه نمایش داده می شود. در این کد ما از خاصیت IsCompleted استفاده کردیم، یکی دیگر از راه ها برای پیاده سازی حالت گفته شده استفاده از متد WaitOn است که در کلاس WaitHandle پیاده سازی شده است. خاصیت AsyncWaitHandle در IAsyncResult شئ ای از نوع WaitHandle بر می گرداند. یکی از مزیت های استفاده از این متد قابلیت مشخص کردن time out برای block کردن thread جاری است، یعنی دیگر نیازی به استفاده از Thread.Sleep نخواهیم داشت:

MathOperation operation = new MathOperation(Add);
Console.WriteLine("Main thread ID:"+Thread.CurrentThread.ManagedThreadId);
var result = operation.BeginInvoke(2, 6, null, null);
while (!result.AsyncWaitHandle.WaitOne(1000,true))
{
    Console.WriteLine("Task in Main method...");
}
var answer = operation.EndInvoke(result);
Console.WriteLine("2 + 6 = {0}", answer);

همانطور که مشاهده می کنید، برای متد WaitOn در حلقه while مقدار 1000 میلی ثانیه یا 1 ثانیه را مشخص کردیم. هر زمان که روند اجرای کد به این دستور می رسد، thread جاری به اندازه 1 ثانیه منتظر تکمیل اجرای متد Add شده و سپس وارد حلقه while می شود. در صورتی که کار متد Add به اتمام برسد، متد WaitOn مقدار true بر میگرداند. در مورد پارامتر دوم WaitOn که مقدار true به آن پاس داده شده در بخش های مرتبط با Synchronization Context صحبت خواهیم کرد.

استفاده از AsyncCallBack

تا اینجا گفتیم که چگونه می توان اجرای همزمان دو Thread را بوسیله Delegate ها پیاده سازی کرد، همچنین با نحوه کنترل دریافت خروجی از Thread های اجرا شده آشنا شدیم و گفتیم که بوسیله IAsyncResult می توان خروجی را دریافت کرد. اگر به خاطر داشته باشید زمان فراخوانی متد BeginInvoke برای دو پارامتر callback و state مقدار null ارسال کردیم. در این قسمت می خواهیم در مورد پارامتر callback صحبت کنیم، این پارامتر که یک Delegate از نوع AsyncCallBack قبول می کند، به ما این امکان را می دهد تا متدی را به متد BeginInvoke پاس دهیم. این delegate زمانی اجرا می شود که روند اجرای متدی که با BeginInvoke فراخوانی شده است به اتمام برسد. متدی که برای callback باید ارسال شود باید به مبتنی بر signature زیر باشد:

void CallBackMethod(IAsyncResult result)
{
     // code for callback
}

مثالی که تا این لحظه بر اساس آن جلو آمدیم را به صورت زیر تغییر می دهیم تا از قابلیت AsynCallBack استفاده کند:

private static bool isDone = false;

static void Main(string[] args)
{
    MathOperation operation = new MathOperation(Add);
    Console.WriteLine("Main thread ID:"+Thread.CurrentThread.ManagedThreadId);
    var result = operation.BeginInvoke(2, 6, new AsyncCallback(CallBack), null);
    while (!isDone)
    {
        Console.WriteLine("Task in Main method...");
        Thread.Sleep(1000);
    }
    var answer = operation.EndInvoke(result);
    Console.WriteLine("2 + 6 = {0}", answer);
}

public static void CallBack(IAsyncResult result)
{
    isDone = true;
}

همانطور که در کد بالا مشاهده می کنید در ابتدا یک فیلد با نام isDone تعریف شده که از آن برای مشخص کردن وضعیت اجرای Thread استفاده می کنیم. زمانی که متد BeginInvoke را فراخوانی می کنیم به عنوان پارامتر سوم شئ ای از نوع AsyncCallback که متد CallBack برای آن مشخص شده ارسال می شود، این متد زمانی فراخوانی می شود که روند اجرای Thread مرتبط با متد Add به اتمام برسد، متد CallBack پس از اجرا مقدار isDone را برابر true قرار می دهد و به همین دلیل از حلقه while تعریف شده در متد Main خارج می شویم.

همانطور که در کد بالا مشاهده می کنید پارامتر ورودی CallBack از نوع IAsyncResult است، یعنی می توان عملیات گرفتن خروجی را در داخل CallBack نیز انجام داد. برای اینکار کد بالا را به صورت زیر تغییر می دهیم:

private static bool isDone = false;

static void Main(string[] args)
{
    MathOperation operation = new MathOperation(Add);
    Console.WriteLine("Main thread ID:"+Thread.CurrentThread.ManagedThreadId);
    operation.BeginInvoke(2, 6, new AsyncCallback(CallBack), null);
    while (!isDone)
    {
        Console.WriteLine("Task in Main method...");
        Thread.Sleep(1000);
    }            
}

public static void CallBack(IAsyncResult result)
{
    AsyncResult asyncRes = (AsyncResult) result;
    var opDelegate = (MathOperation) asyncRes.AsyncDelegate;
    var answer = opDelegate.EndInvoke(result);
    Console.WriteLine("2 + 6 = {0}", answer);
    isDone = true;
}

تغییراتی که در کد بالا دادیم به ترتیب:

  1. ابتدا در متد CallBack پارامتر ورودی result را به کلاس AsyncResult تبدیل کردیم، کلاس AsyncResult اینترفیس IAsyncResult را پیاده سازی کرده است و علاوه بر اعضاء این اینترفیس یکسری اعضاء دیگر دارد از جمله AsyncDelegate که شئ delegate فراخوانی شده را برای ما بر می گرداند.
  2. در قدم بعدی AsyncDelegate را به MathOperation تبدیل کردیم.
  3. در انتها عملیاتی که در متد Main برای گرفتن خروجی نوشته بودیم را به متد CallBack منتقل کردیم تا بتوانیم خروجی متد Add را گرفته و در پنجره Console نمایش دهیم.
  4. در انتها مقدار isDone را برابر true قرار دادیم تا اطلاع دهیم عملیات اجرای متد به پایان رسیده است.

ارسال و دریافت داده های دلخواه بین Thread ها

در خاتمه این قسمت آموزشی با پارامتر چهارم متد BeginInvoke، یعنی state آشنا می شویم. بوسیله این پارامتر می توان یک مقدار یا شئ دلخواه را به Delegate ارسال و از آن بوسیله IAsyncResult استفاده کرد. برای مثال، می خواهیم زمانی که Delegate را فراخوانی می کنیم یک رشته را به delegate پاس داده و در متد CallBack آن را نمایش دهیم. کد نوشته شده را به صورت زیر تغییر می دهیم:

private static bool isDone = false;

static void Main(string[] args)
{
    MathOperation operation = new MathOperation(Add);
    Console.WriteLine("Main thread ID:"+Thread.CurrentThread.ManagedThreadId);
    operation.BeginInvoke(2, 6, new AsyncCallback(CallBack), "This is state passed to thread from Main Method!!");
    while (!isDone)
    {
        Console.WriteLine("Task in Main method...");
        Thread.Sleep(1000);
    }            
}

public static void CallBack(IAsyncResult result)
{
    Console.WriteLine("State: " + result.AsyncState);
    AsyncResult asyncRes = (AsyncResult) result;
    var opDelegate = (MathOperation) asyncRes.AsyncDelegate;
    var answer = opDelegate.EndInvoke(result);
    Console.WriteLine("2 + 6 = {0}", answer);
    isDone = true;
}

در متد و زمان فراخوانی delegate بوسیله BeginInvoke به عنوان پارامتر چهارم یک رشته را به عنوان state ارسال کرده و در CallBack بوسیله خاصیت AsyncState توانستیم state ارسال شده را گرفته و در خروجی نمایش دهیم.
تا این قسمت شما یاد گرفتید که چگونه در زبان دات نت می توان با کمک Delegate ها اقدام به اجرای کدها در یک Thread جداگانه کرد. در قسمت های بعدی آموزش با نحوه استفاده از کلاس Thread که در فضای نام System.Threading قرار دارد بیشتر آشنا خواهیم شد.

منبع


قسمت اول آموزش-برنامه نویسی Asynchronous – آشنایی با Process ها، Thread ها و AppDomain ها

قسمت دوم آموزش- آشنایی با ماهیت Asynchronous در Delegate ها

قسمت سوم آموزش-آشنایی با فضای نام System.Threading و کلاس Thread

قسمت چهارم آموزش- آشنایی با Thread های Foreground و Background در دات نت

قسمت پنجم آموزش- آشنایی با مشکل Concurrency در برنامه های Multi-Threaded و راهکار های رفع این مشکل

قسمت ششم آموزش- آشنایی با کلاس Timer در زبان سی شارپ

قسمت هفتم آموزش-آشنایی با CLR ThreadPool در دات نت

قسمت هشتم آموزش- مقدمه ای بر Task Parallel Library و کلاس Parallel در دات نت

قسمت نهم آموزش- برنامه نویسی Parallel:آشنایی با کلاس Task در سی شارپ

قسمت دهم آموزش-برنامه نویسی Parallel در سی شارپ :: متوقف کردن Task ها در سی شارپ – کلاس CancellationToken

قسمت یازدهم آموزش- برنامه نویسی Parallel در سی شارپ :: کوئری های Parallel در LINQ

قسمت دوازدهم آموزش- آشنایی با کلمات کلیدی async و await در زبان سی شارپ

قسمت سیزدهم آموزش- استفاده از متد WhenAll برای اجرای چندین Task به صورت همزمان در سی شارپ

 

 

 

کار با Thread ها در زبان سی شارپ – آشنایی با فضای نام System.Threading و کلاس Thread

تا اینجا متوجه شدیم که چگونه می توان با کمک Delegate ها کدها را در یک Thread جداگانه و به صورت Asynchrnonous اجرا کرد. در ادامه مباحث مرتبط با برنامه نویسی Asynchronous به سراغ فضای نام System.Threading می رویم. این فضای نام شامل یکسری کلاس است که روند نوشتن برنامه Multi-Threaded را آسان می کند. کلاس های زیادی در این فضای نام وجود دارد که هر یک استفاده خاص خودش را دارد. در زیر با توضیح اولیه برخی از کلاس های این فضای نام آشنا می شویم:

  1. Interlocked: از این کلاس برای اجرای عملیات های atomic یا Atomic Operations بر روی متغیرهایی که در بین چندین Thread به اشتراک گذاشته شدند استفاده می شود.
  2. Monitor: از این کلاس برای پیاده سازی Synchronization بر روی اشیاء ای که Thread به آن دسترسی دارند استفاده می شود. در سی شارپ کلمه کلیدی lock در پشت زمینه از مکانیزم Monitor برای Synchronization استفاده می کنید که در بخش های بعدی با این تکنیک بیشتر آشنا می شویم.
  3. Mutex: از این کلاس برای اعمال Synchronization بین AppDomain ها استفاده می شود.
  4. ParameterizedThreadStart: این delegate به thread ها این اجازه را می دهد تا متدهایی پارامتر ورودی دارند را فراخوانی کند.
  5. Semaphor: از این کلاس برای محدود کردن تعداد Thread هایی که می توانند به یک Resource دسترسی داشته باشند استفاده می شود.
  6. Thread: به ازای هر Thread ایجاد شده برای برنامه باید یک کلاس از نوع Thread ایجاد کرد. در حقیقت کلاس Thread نقش اصلی را در ایجاد و استفاده از Thread ها دارد.
  7. ThreadPool: بوسیله این کلاس می توان به Thread-Pool مدیریت شده توسط خود CLR دسترسی داشت.
  8. ThreadPriority: بوسیله این enum می توان درجه اهمیت یک Thread را مشخص می کند.
  9. ThreadStart: از این delegate برای مشخص کردن متدی که در Thread باید اجرا شود استفاده می شود. این delegate بر خلاف ParameterizedThreadStart پارامتری قبول نمیکند.
  10. ThreadState: بوسیله این enum می توان وضعیت جاری یک thread را مشخص کرد.
  11. Timer: بوسیله کلاس Timer می توان مکانیزمی پیاده کرد که کدهای مورد نظر در بازه های زمانی خاص، مثلاً هر 5 ثانیه یکبار و در یک Thread مجزا اجرا شوند.
  12. TimerCallBack: از این delegate برای مشخص کردن کدی که داخل timer باید اجرا شود استفاده می شود.

کلاس System.Threading.Thread

کلاس Thread اصلی ترین کلاس موجود در فضای نام System.Threading است. از یک کلاس برای دسترسی به Thread هایی که در روند اجرای یک AppDomain ایجاد شده اند استفاده می شود. همچنین بوسیله این کلاس می تواند Thread های جدید را نیز ایجاد کرد. کلاس Thread شامل یکسری متد و خصوصیت است که در این قسمت می خواهیم با آن ها آشنا شویم. ابتدا به سراغ خصوصیت CurrentThread که یک خصوصیت static در کلاس Thread است می رویم. بوسیله این خصوصیت می توان اطلاعات Thread جاری را بدست آورد. برای مثال در صورتی که در متد Main از این خصوصیت استفاده شود می توان به اطلاعات مربوط به Thread اصلی برنامه دسترسی داشت یا اگر برای یک متد Thread جداگانه ای ایجاد شود، در صورت استفاده از این خصوصیت در بدنه متد به اطلاعات Thread ایجاد شده دسترسی خواهیم داشت. در ابتدا با یک مثال می خواهیم اطلاعات Thread اصلی برنامه را بدست آوریم:

var primaryThread = Thread.CurrentThread;
primaryThread.Name = "PrimaryThread";
Console.WriteLine("Thread Name: {0}", primaryThread.Name);
Console.WriteLine("Thread AppDomain: {0}", Thread.GetDomain().FriendlyName);
Console.WriteLine("Thread Context Id: {0}", Thread.CurrentContext.ContextID);
Console.WriteLine("Thread Statred: {0}", primaryThread.IsAlive);
Console.WriteLine("Thread Priority: {0}", primaryThread.Priority);
Console.WriteLine("Thread State: {0}", primaryThread.ThreadState);

دقت کنید در ابتدا با به Thread یک نام دادیم. در صورتی که نامی برای Thread انتخاب نشود خصوصیت Name مقدار خالی بر میگرداند. مهمترین مزیت تخصیص نام برای Thread راحت تر کردن امکان debug کردن کد است. در Visual Studio پنجره ای وجود دارد به نام Threads که می توانید در زمان اجرای برنامه از طریق منوی Debug->Windows->Threads به آن دسترسی داشته باشید. در تصویر زیر نمونه ای از این پنجره را در زمان اجرا مشاهده می کنید:

آشنایی با فضای نام System.Threading و کلاس Thread

اما بریم سراغ موضوع اصلی، یعنی ایجاد Thread و اجرای آن. در ابتدا در مورد مراحل ایجاد یک Thread صحبت کنیم، معمولاً برای ایجاد یک Thread مراحل زیر را باید انجام دهیم:

  1. در ابتدا باید متدی ایجاد کنیم که وظیفه آن انجام کاری است که قرار است در یک Thread جداگانه انجام شود.
  2. در مرحله بعد باید یکی از delegate های ParameterizedThreadStart برای متدهایی که پارامتر ورودی دارند یا ThreadStart برای متدهای بدون پارامتر را انتخاب کرده و یک شئ از آن ایجاد کنیم که به عنوان سازنده متد مورد نظر به آن پاس داده می شود.
  3. از روی کلاس Thread یک شئ جدید ایجاد کرده و به عنوان سازنده شئ ای که از روی delegate های گفته شده در مرحله 2 ساختیم را به آن ارسال کنیم.
  4. اطلاعات و تنظیمات اولیه مورد نظر برای Thread مانند Name یا Priority را برای آن ست کنیم.
  5. متد Start را در کلاس Thread را برای شروع کار Thread فراخوانی کنیم. با این کار متدی که در مرحله 2 مشخص کردیم در یک Thread جداگانه اجرا می شود.

دقت کنید در مرحله 2 می بایست بر اساس signature متدی که قصد اجرای آن در thread جداگانه را داریم، delegate مناسب انتخاب شود. همچنین ParameterizedThreadStart پارامتری که به عنوان ورودی قبول می کند از نوع Object است، یعنی اگر می خواهید چندین پارامتر به آن ارسال کنید می بایست حتماً یک کلاس یا struct ایجاد کرده و آن را به عنوان ورودی به کلاس Start ارسال کنید. با یک مثال ساده که از ThreadStart برای اجرای Thread استفاده می کند شروع می کنیم:

static void Main(string[] args)
{
    ThreadStart threadStart = new ThreadStart(PrintNumbers);
    Thread thread = new Thread(threadStart);
    thread.Name = "PrintNumbersThread";
    thread.Start();
    while (thread.IsAlive)
    {
        Console.WriteLine("Running in primary thread...");
        Thread.Sleep(2000);
    }
    Console.WriteLine("All done.");
    Console.ReadKey();
}

public static void PrintNumbers()
{
    for (int counter = 0; counter < 10; counter++)
    {
        Console.WriteLine("Running from thread: {0}", counter + 1);
        Thread.Sleep(500);
    }
}

در ابتدا متدی تعریف کردیم با نام PrintNumbers که قرار است در یک Thread مجزا اجرا شود. همانطور که مشاهده می کنید این متد نه پارامتر ورودی دارد و نه مقدار خروجی، پس از ThreadStart استفاده می کنیم. بعد از ایجاد شئ از روی ThreadStart و ایجاد Thread، نام Thread را مشخص کرده و متد Start را فراخوانی کردیم. به حلقه while ایجاد شده دقت کنید، در این حلقه بوسیله خصوصیت IsAlive گفتیم تا زمانی که Thread ایجاد شده در حال اجرا است کد داخل while اجرا شود. همچنین بوسیله متد Sleep در متد Main و متد PrintNumbers در عملیات اجرا برای Thread های مربوط به متد تاخیر ایجاد کردیم. بعد اجرای کد بالا خروجی زیر نمایش داده می شود:

Running in primary thread...
Running from thread: 1
Running from thread: 2
Running from thread: 3
Running from thread: 4
Running in primary thread...
Running from thread: 5
Running from thread: 6
Running from thread: 7
Running from thread: 8
Running in primary thread...
Running from thread: 9
Running from thread: 10
All done.

در قدم بعدی فرض کنید که قصد داریم بازه اعدادی که قرار است در خروجی چاپ شود را به عنوان پارامتر ورودی مشخص کنیم، در اینجا ابتدا یک کلاس به صورت زیر تعریف می کنیم:

public class PrintNumberParameters
{
    public int Start { get; set; }
    public int Finish { get; set; }
}

در قدم بعدی کلاس PrintNumbers را به صورت زیر تغییر می دهیم:

public static void PrintNumbers(object data)
{
    PrintNumberParameters parameters = (PrintNumberParameters) data;
    for (int counter = parameters.Start; counter < parameters.Finish; counter++)
    {
        Console.WriteLine("Running from thread: {0}", counter);
        Thread.Sleep(500);
    }
}

همانطور که مشاهده می کنید، پارامتر ورودی PrintNumbers از نوع object است و در بدنه ورودی را به کلاس PrintNumberParameters تبدیل کرده و از آن استفاده کردیم. در مرحله بعد متد Main را باید تغییر داده و به جای ThreadStart از ParameterizedThreadStart استفاده کنیم، همچنین به عنوان پارامتر ورودی برای متد Start شئ ای از PrintNumberParameters ایجاد کرده و با عنوان پارامتر به آن ارسال می کنیم:

ParameterizedThreadStart threadStart = new ParameterizedThreadStart(PrintNumbers);
Thread thread = new Thread(threadStart);
thread.Name = "PrintNumbersThread";
thread.Start(new PrintNumberParameters() {Start = 5, Finish = 13});
while (thread.IsAlive)
{
    Console.WriteLine("Running in primary thread...");
    Thread.Sleep(2000);
}
Console.WriteLine("All done.");
Console.ReadKey();

با اعمال تغییرات ذکر شده و اجرای کد، اعداد بر اساس بازه مشخص شده در خروجی چاپ می شوند. در این قسمت از مطلب مربوط به Thread ها با نحوه ایجاد و استفاده از Thread ها آشنا شدیم. در قسمت های بعدی به مباحث دیگری در مورد Thread ها خواهیم پرداخت.

منبع


قسمت اول آموزش-برنامه نویسی Asynchronous – آشنایی با Process ها، Thread ها و AppDomain ها

قسمت دوم آموزش- آشنایی با ماهیت Asynchronous در Delegate ها

قسمت سوم آموزش-آشنایی با فضای نام System.Threading و کلاس Thread

قسمت چهارم آموزش- آشنایی با Thread های Foreground و Background در دات نت

قسمت پنجم آموزش- آشنایی با مشکل Concurrency در برنامه های Multi-Threaded و راهکار های رفع این مشکل

قسمت ششم آموزش- آشنایی با کلاس Timer در زبان سی شارپ

قسمت هفتم آموزش-آشنایی با CLR ThreadPool در دات نت

قسمت هشتم آموزش- مقدمه ای بر Task Parallel Library و کلاس Parallel در دات نت

قسمت نهم آموزش- برنامه نویسی Parallel:آشنایی با کلاس Task در سی شارپ

قسمت دهم آموزش-برنامه نویسی Parallel در سی شارپ :: متوقف کردن Task ها در سی شارپ – کلاس CancellationToken

قسمت یازدهم آموزش- برنامه نویسی Parallel در سی شارپ :: کوئری های Parallel در LINQ

قسمت دوازدهم آموزش- آشنایی با کلمات کلیدی async و await در زبان سی شارپ

قسمت سیزدهم آموزش- استفاده از متد WhenAll برای اجرای چندین Task به صورت همزمان در سی شارپ

در دوربین های مداربسته ی آنالوگ، رزولوشن با TV Line اندازه گیری می شود. در اکثر دوربین های آنالوگ جدید، TVL بین 420 تا 700 متغیر است. اگر چه 420TVL پایین ترین محسوب می شود، سیستم های امنیت تصویری وجود دارد که لازم است فقط یک فاصله ی کوتاه پوشش داده شود و در اینگونه موارد دوربین های 420TVL خوب کار می کنند.

دوربین های 600TVL به بالا، تصاویر دقیق تر و کنتراست بالایی دارند. اگر نیاز به تصویر صاف و دقیق دارید، دوربین های بالای 600TVL را انتخاب کنید هر چند فاکتور های مهم دیگری مانند میزان لوکس، WDR و … در تصویر دریافتی از دوربین تاثیر گذار هستند.

وقتی دوربین ها تصاویر را دریافت می کنند، آن را از طریق کابل به DVR می فرستند. در DVR، تصاویر از حالت آنالوگ به حالت دیجیتال تبدیل می شوند تا هم روی هارد ذخیره شوند و هم از طریق مانیتور نمایش داده شوند. این مهمترین بخش زنجیره ی دریافت و ذخیره ی تصاویر در دوربین های مدار بسته است.

بهترین دوربین ها با بالاترین کیفیت هم در صورتی که از یک DVR با توانایی تبدیل پایین استفاده شود نمی تواند تصویر خوبی به شما بدهد. DVR ها دو مدل رزولوشن پرکاربرد دارند : D1 , Cif

Cif سایز 320*240 پیکسل به شما می دهد و D1 سایز 720*480 و همانطور که مشخص است، D1 چهار برابر بزرگتر از Cif است پس تصویر کمتر فشرده می شود و در نتیجه جزئیات بیشتری را نشان می دهد.

برای انتخاب DVR، رزولوشن ضبط تصاویر را در نظر داشته باشید. بعضی از DVR ها هر دو گزینه را دارند ولی معمولا تعداد فریم بر ثانیه را پایین می آورد تا تصاویر پر کیفیت تری ارائه دهد.

 

رزولوشن در دوربین های مدار بسته

 

دوربین های آی پی (IP) قابلیت این را دارند که تصاویر با رزولوشن خیلی خیلی بالاتر داشته باشند. اولین چیزی که باید در مورد دوربین های آی پی بدانید این است که تصاویر را به صورت دیجیتال دریافت می کند به همین خاطر نیازی به تبدیل یا فشرده سازی وجود ندارد. دومین مسئله این است که دوربین های آی پی تصاویر را از طریق کابل های Ca45 یا Ca46 منتقل می کند که قابلیت گذردهی خیلی بیشتری دارند.

دوربین های 1.3 مگاپیکسل معمولا کوچکترین رزولوشن دوربین های آی پی است( البته رزولوشن های پایین تر هم وجود دارد) که خیلی خیلی بزرگتر و با کیفیت تر از هر دوربین آنالوگ دیگری است.

با این وجود اگر حتی تصاویر 1280*1024 پیکسل هم برایتان کوچک است می توانید از دوربین های 3 مگا پیکسل که رزولوشن 2048*1536 پیکسل دارند استفاده کنید یا حتی اگر این هم کم است می توانید دوربین های 5 مگا پیکسل (1944*2592) استفاده کنید.

خلاصه اینکه با دوربین های آنالوگ و استفاده از DVR شما نهایت تصاویر D1, 4Cif, 2Cif ,Cif خواهید داشت که ممکن است بتوانند نیازهای شما را برآورده کنند. ولی اگر نیاز به رزولوشن بالاتری دارید باید دوربین های آی پی را در نظر بگیرید و از آنها استفاده کنید.

 

منبع : http://www.elmcctv.ir

 

موازی سازی(Parallelism) چیست؟

ﺷﺮﮐﺖ ﻫﺎی ﺗﻮﻟﯿﺪ ﮐﻨﻨﺪه ﭘﺮدازﺷﮕﺮ ﺑﺮای اﻓﺰاﯾﺶ ﺳﺮﻋﺖ ﭘﺮدازﻧﺪه ﻣﺠﺒﻮر ﺑﻪ ﺑﺎﻻﺑﺮدن ﻓﺮﮐﺎﻧﺲ ﭘﺮدازﺷﮕﺮ ﺑﻮدﻧﺪ. ﯾﮏ راه اﻓﺰاﯾﺶ وﻟﺘﺎژ ﻣﺼﺮﻓﯽ ﭘﺮدازﻧﺪه ﺑﻮد ﮐﻪ دارای ﻧﻘﺎط ﺿﻌﻔﯽ ﻣﺎﻧﻨﺪ اﻓﺰاﯾﺶ دﻣﺎ و اﻓﺰاﯾﺶ ﻣﺼﺮف ﺑﺎﻃﺮی ﻧﯿﺰ ﺑﻮد. از ﻃﺮﻓﯽ ﺗﻮﻟﯿﺪ ﮐﻨﻨﺪﮔﺎن ﭘﺮدازﻧﺪه ﺑﻪ ﮐﻤﮏ ﺑﺮﻧﺎﻣﻪ ﻧﻮﯾﺴﺎن ﭘﯽ ﺑﻪ ﺑﯿﮑﺎری زﯾﺎد ﭘﺮدازﺷﮕﺮﻫﺎ در زﻣﺎن ﺳﻮﯾﭻ ﮐﺮدن ﻓﺮاﯾﻨﺪ ﻫﺎ و ﻧﺦ ﻫﺎ ﺷﺪﻧﺪ ﮐﻪ ﺣﺪود ﻧﯿﻤﯽ از زﻣﺎن ﭘﺮدازش را ﺑﻪ ﻫﺪر ﻣﯽ داد. ﺑﺮاي ﺟﺒﺮان ﺣﺎﻓﻈﻪ ﮐﺶ را ﮔﺴﺘﺮش دادﻧﺪ اﻣﺎ ﺑﻪ دﻟﯿﻞ ﮔﺮان ﺑﻮدﻧﺶ ﺑﺎز دﭼﺎر ﻣﺤﺪودﯾﺖ ﺑﻮدﻧﺪ. ﺑﻨﺎﺑﺮاﯾﻦ ﭘﺮدازﻧﺪه ﻫﺎﯾﯽ ﺗﻮﻟﯿﺪ ﮐﺮدﻧﺪ ﮐﻪ ﺑﺘﻮاﻧﺪ ﭘﺮازش ﻣﻮازي را (در اﺑﺘﺪا) در دو ﻫﺴﺘﻪ ﺑﻪ اﺟﺮا ﺑﺮﺳﺎﻧﻨﺪ.

ﻧﺎم اﯾﻦ ﻫﺴﺘﻪ ﻫﺎ ﻫﺴﺘﻪ ﻫﺎی ﺳﺨﺖ اﻓﺰاری ﯾﺎ ﻓﯿﺰﯾﮑﯽ ﮔﺬاﺷﺘﻨﺪ. اﻧﺪك زﻣﺎﻧﯽ ﺑﻌﺪ ﻓﻨﺎوری ای ﺑﺮای رﺳﯿﺪن ﺑﻪ ﭘﺮدازش ﻣﻮازی اﻣﺎ در ﺳﻄﺢ ﻣﺤﺪود Hyper ﺗﺮی و ارزان ﺗﺮ ﺑﺎ ﻧﺎم اﺑﺮ ﻧﺨﯽ ﯾﺎ Hyper-Threading اراﺋﻪ ﮐﺮدﻧﺪ و ﻧﺎم آن را ﻫﺴﺘﻪ ﻫﺎی ﻣﻨﻄﻘﯽ ﯾﺎ ﻧﺦ ﻫﺎی ﺳﺨﺖ اﻓﺰاری ﮔﺬاﺷﺘﻨﺪ. ﺣﺎل ﻧﻮﺑﺖ ﺑﺮﻧﺎﻣﻪ ﻧﻮﯾﺴﺎن ﺑﻮد ﺗﺎ ﺑﺮﻧﺎﻣﻪ ﻫﺎی ﺑﺮاي اﺳﺘﻔﺎده از اﯾﻦ ﻓﻨﺎوری ﻫﺎی ﻧﻮﯾﻦ ﺑﻨﻮﯾﺴﻨﺪ. ﺑﺮﻧﺎﻣﻪ ﻧﻮﯾﺴﯽ ﻣﻮازی ﻋﻨﻮاﻧﯽ اﺳﺖ، ﮐﻪ ﻣﻮﺿﻮﻋﯽ ﮔﺴﺘﺮده در دﻧﯿﺎي ﻧﺮم اﻓﺰار اﯾﺠﺎد ﮐﺮده است.

روش موازی سازی(Parallelism(

ﺳﺎده ﺗﺮﯾﻦ ﺷﯿﻮه ﻣﻮازي ﺳﺎزي در ﻗﺎﻟﺐ Task ها ﺻﻮرت ﻣﯽ ﮔﯿﺮد، در ﻫﺮ Task ﺗﺎﺑﻊ ﯾﺎ ﻗﻄﻌﻪ ﮐﺪي ﻧﻮﺷﺘﻪ ﻣﯽ ﺷﻮد و ﺳﭙﺲ ﺑﻮﺳﯿﻠﻪ Delegate اي ﮐﻪ ﮐﺎر ﻣﺪﯾﺮﯾﺖ Task ﻫﺎ را ﺑﺮ ﻋﻬﺪه دارد اﯾﻦ  Task ﻫﺎ ﺑﺼﻮرت ﻣﻮازی ﺑﺴﺘﻪ ﺑﻪ ﻫﺴﺘﻪ ﻫﺎی ﻣﻨﻄﻘﯽ در دﺳﺘﺮس اﺟﺮا ﻣﯽ ﺷﻮﻧﺪ. روش ﻫﺎی ﺑﺴﯿﺎری ﺑﺮاي ﻣﻮازی ﺳﺎزی وﺟﻮد دارد ﻣﺎﻧﻨﺪ اﺳﺘﻔﺎده از ﮐﻼس Parallel.For یا Parallel.ForEach ﮐﻪ در ﺟﺎي ﺧﻮد ﮐﺎرﺑﺮد ﻫﺎی ﻣﺨﺘﺺ ﺑﻪ ﺧﻮدﺷﺎن را دارﻧﺪ. همیشه الگورﯾﺘﻢ ﻫﺎی ﺗﺮﺗﯿﺒﯽ را ﻧﻤﯽ ﺗﻮان ﺑﻪ الگورﯾﺘﻤﯽ ﻣﻮازي ﺗﺒﺪﯾﻞ ﮐﺮد ﭼﺮا ﮐﻪ ﮐﺪ ﻫﺎی ﺗﺮﺗﯿﺒﯽ ای ﻫﺴﺘﻨﺪ ﮐﻪ اﺟﺮاي ﮐﺪ ﻫﺎي دﯾﮕﺮ ﻧﯿﺎز ﺑﻪ ﺗﮑﻤﯿﻞ ﺷﺪن آن ﻫﺎ دارد. ﺑﺴﺘﻪ ﺑﻪ الگورﯾﺘﻢ درﺻﺪی از آن را ﻣﯽ ﺗﻮان ﻣﻮازی ﮐﺮد. ﻗﺒﻞ از ﻣﻮازی ﺳﺎزی ﺑﺎﯾﺪ ﻣﻮازی ﺳﺎزی را در ذﻫﻨﺘﺎن ﻃﺮاﺣﯽ ﮐﻨﯿﺪ.


Parallel Programming  یا برنامه نویسی موازی یعنی تقسیم یک مسئله به مسائل کوچکتر و سپردن آن ها به واحد های جداگانه برای پردازش کردن.این مسائل کوچک به صورت همزمان شروع به اجرا می کنند. Parallel Programming وظیفه یا Task را به اجزا مختلفی تقسیم می کند.

فرم های مختلفی از Parallel وجود دارد .مانند bit-level ،  instruction-level، data ، taskدر این آموزش راجع به Data Parallelism و Task Parallelism بحث خواهیم کرد.

تصور کنید هسته CPUمتشکل از چندین ریزپردازنده است که همه این ها به حافظه اصلی دسترسی دارند.هر کدام از این ریزپردازنده ها قسمتی از مسئله را حل می کنند.

Data Parallelism

این مورد بر روی توزیع دیتا در نقاط مختلف تمرکز می کند.یعنی داده را به بخش های مختلفی می شکند.هر کدام از این بخش ها به Thread جداگانه ای برای پردازش داده می شود.

Task Parallelism

این مفهوم وظایف یا Taskها را به بخش هایی شکسته و هر کدام را به یک Thread جهت پردازش می دهد.

در پروژه ای که به صورت ضمیمه این مقاله می باشد (در پروژه DataParallisem)به سه صورت مختلف وظیفه یا Task تعریف شده است.

1-به صورت Function

2- به صورت Delegate

3-به صورت لامبدا

کد این قسمت به صورت زیر می باشد

using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Threading.Tasks;
  
namespace TPL_part_1_creating_simple_tasks
{
    class Program
    {
        static void Main(string[] args)
        {
            //Action delegate
            Task task1 = new Task(new Action(HelloConsole));
 
            //anonymous function
            Task task2 = new Task(delegate
            {
                HelloConsole();
            });
             
            //lambda expression
                Task task3 = new Task(() = > HelloConsole());                 
             
            task1.Start();
            task2.Start();
            task3.Start();
             
            Console.WriteLine("Main method complete. Press any key to finish.");
            Console.ReadKey();
        }
        static void HelloConsole()
        {
            Console.WriteLine("Hello Task");
        }
    }
}

 

بعد از اجرا، هر کدام از Task ها اجرا شده البته به صورت همزمان و کارهای محوله به آنها را انجام میدهند.

آموزش موازی سازی در سی شارپ

 

در این آموزش بر روی مفهوم Data Parallelism تمرکز خواهیم کرد.توسط Data Parallelism عملیات یکسانی بر روی المانهای یک مجموعه یا آرایه به صورت همزمان انجام خواهد شد. که در فضای نام System.Threading.Tasks.Parallel قرار دارد. روش اصلی برای انجام Data Parallelismنوشتن یک تابع است که یک حلقه ساده بدون Thread دارد .

public static void DataOperationWithForeachLoop()
       {
           var mySource = Enumerable.Range(0, 1000).ToList();
           foreach (var item in mySource)
           {
               Console.WriteLine("Square root of {0} is {1}", item, item * item);
           }
       }

 

خروجی برنامه را در زیر می بینید

آموزش موازی سازی در سی شارپ

 

عملیات Data Parallelism را می توان با یک حلقه foreach موازی هم انجام داد.

public static void DataOperationWithDataParallelism()
        {
            var mySource = Enumerable.Range(0, 1000).ToList();
            Parallel.ForEach(mySource, values = > CalculateMyOperation(values));
        }
 
        public static  void CalculateMyOperation(int values)
        {
            Console.WriteLine("Square root of {0} is {1}", values, values * values);
        }

بعد از اجرا شکل زیر را خواهید دید.

آموزش موازی سازی در سی شارپ

 

در این کد در داخل حلقه Foreach  یک تابع Delegate قرار دادیم در این تابع به ازای هر تکرار حلقه بر روی مجموعه تابعی که درون Delegate فراخوانی کرده ایم اجرا خواهد شد.

Data Parallelism توسط PLINQ

PLINQ به معنای Parallel LINQ است .این نسخه از لینک جهت پیاده سازی لینک بر روی پردازنده های چند هسته ای نوشته شده است.

توسط لینک می توان اطلاعات را از چندین منبع بازیابی کرد.و در نهایت این نتایج با هم ترکیب می شوند تا نتیجه نهایی Query به دست آید.اما اگر از PLINQاستفاده کنیم این دستورات به جای اینکه پشت سر هم اجرا شوند به صورت موازی اجرا می شوند.

برای این که از PLINQ استفاده کرد فقط کافی است که در انتهای عبارت لینک از AsParallel استفاده کنیم.به کد زیر توجه کنید

public static void DataOperationByPLINQ()
    {
        long mySum = Enumerable.Range(1, 10000).AsParallel().Sum();
        Console.WriteLine("Total: {0}", mySum);
    }

بعد از اجرا شکل زیر را خواهید دید

آموزش موازی سازی در سی شارپ

 

برای به دست آوردن اعداد فرد در این مجموعه توسط Plinq از کد زیر استفاده می کنیم

public static void ShowEvenNumbersByPLINQ()
      {
          var numers = Enumerable.Range(1, 10000);
          var evenNums = from number in numers.AsParallel()
                         where number % 2 == 0
                         select number;
 
          Console.WriteLine("Even Counts :{0} :", evenNums.Count());
      }

پس از اجرا شکل زیر را خواهید دید

آموزش موازی سازی در سی شارپ

 

توسط متد Parallel.Invoke() می توانید چندین متد را مانند شکل زیر به صورت همزمان اجرا کنید.

Parallel.Invoke(    
() = > Method1(mycollection),    
() = > Method2(myCollection1, MyCollection2),    
() = > Method3(mycollection));

 MaxDegreeOfParallelism

ماکزیمم تعداد پردازش های موازی را مشخص می کند در کد زیر و در داخل Foreach در پارامتر دوم ماکزیمم تعداد پردازش های موازی مشخص شده اشت.

loopState.Break()

توسط این کد به Thread هایی که پردازش آنها طول کشیده اجازه می دهیم که بعدا Break شوند.به کد زیر توجه کنید.

var mySource = Enumerable.Range(0, 1000).ToList();    
int data = 0;    
Parallel.ForEach(    
    mySource,    
    (i, state) = >    
    {    
        data += i;    
        if (data  >  100)    
        {    
            state.Break();    
            Console.WriteLine("Break called iteration {0}. data = {1} ", i, data);    
        }    
    });    
Console.WriteLine("Break called data = {0} ", data);    
Console.ReadKey();

منبع

 


فایل ضمیمه این آموزش

TPL_part_1_creating_simple_tasks

رمز فایل: behsanandish.com


دانلود کتاب آموزش برنامه نویسی موازی با #C

Parallel Programming In Csharp

رمز فایل: behsanandish.com

 

مقدمه

حذف نویز تصاویر _ گروهی از محققان سیستمی را توسعه داده اند که با استفاده از هوش مصنوعی و بدون نیاز به عکس های واضح از منبع، نویز تصاویر را از بین می برد.

شرح خبر

این گروه متشکل از محققان انویدیا، MIT و دانشگاه آلتو در توسعه این سیستم از یادگیری عمیق بهره برده اند که بر خلاف روش های قبلی نیازی به مشاهده نمونه های کامل از تصویر مورد نظر داشته و تنها با استفاده از داده های ناقص یا دو تصویر غیر واضح به افزایش کیفیت تصاویر می پردازد. علاوه بر این نتیجه نهایی افزایش کیفیت، حذف متون یا اصلاح تصویر نسبت به روش های قبلی به مراتب بهتر است.

یادگیری عمیق گونه ای از یادگیری ماشینی است که در آن سیستم با کمک هوش مصنوعی نحوه بازیابی تصاویر دارای نویز از طریق کنار هم قرار دادن تصاویر، متون یا ویدیوها را فرا می گیرد. یکی دیگر از قابلیت های جالب توجه سیستم جدید افزایش کیفیت تصاویر در عرض چند میلی ثانیه است.
مبنای کار هوش مصنوعی در این سیستم بر شبکه عصبی استوار است که با استفاده از تصاویر دارای نویز آموزش دیده است. در این روش هوش مصنوعی علی رغم عدم نیاز به تصاویر واضح از منبع باید دوبار تصویر را مشاهده کند.

آزمایشات این گروه نشان داده که از تصاویر تخریب شده از طریق نویزهایی نظیر «گاوسی افزایشی»، «پواسون» یا ترکیب آنها می توان برای تولید تصاویر بهینه ای استفاده کرد که کیفیت آن‌ها با تصاویر بازیابی‌ شده از عکس های بدون مشکل تقریبا برابر است.
کاربردهای علمی این سیستم مبتنی بر یادگیری عمیق شامل زمینه های پزشکی است که در آن می توان کیفیت اسکن های MRI و تصاویر دیگر را به شکل چشمگیری افزایش داد.

چند ماه قبل نیز تیم تحقیقاتی انستیتوی «ماکس پلانک» به رهبری دکتر مهدی سجادی، الگوریتمی را توسعه داده بودند که با بهره گیری از هوش مصنوعی وضوح تصاویر بی کیفیت را تا حد زیادی بهبود می بخشید.

کد برنامه تشخیص و شمارش خودروهای در حال حرکت در زبان Matlab

در این پروژه سعی داریم با استفاده از روش «رهگیری اهداف چندگانه» اقدام به شناسایی اشیاء در حال حرکت نموده و آنها را شمارش نمائیم. این برنامه می تواند جهت شمارش تعداد خودروهای عبوری، تعداد افراد در حال تردد و… مورد استفاده قرار بگیرد. این الگوریتم توسط شرکت mathwork پیاده سازی گردیده و جهت استفاده علاقه مندان ادامه ارائه می گردد.

باتوجه به توضیحات کامل این پروژه از ذکر توضیحات اضافه اجتناب میکنیم و فقط شرحی مختصر از عماکرد برنامه ارائه می نماییم. ابتدا با استفاده از چند فریم به عنوان نمونه، زمینه(background) را شناسایی می کنیم و سپس به اقدام به تشخیص آبجکت های(foreground) می نماییم. سپس با استفاده از روش کالمن (Kalman) اقدام به رهگیری آبجکت هایی که از مقداری مشخص (اصلاحا blob) بزرگتر هستند می نمائیم.  در ادامه آبجکت ها را رهگیری می کنیم تا هنگامی که از صفحه خارج شوند. نکته جالب توجه این هست که اگر آبجکتی موقتا ناپیدا شود(مثلا زیر پل یا درخت قرار بگیرد) به عنوان «Predicted» برچسب خورده و پس از پیدا شدن مجددا به عنوان همان آبجکت قبلی شناسایی می شود.

 

تشخیص و شمارش خودروهای در حال حرکت

سورس کد برنامه تشخیص و شمارش خودرو:

 

%% Multiple Object Tracking Tutorial
% This example shows how to perform automatic detection and motion-based
% tracking of moving objects in a video. It simplifies the example
% <matlab:helpview(fullfile(docroot,'toolbox','vision','vision.map'),'MotionBasedMultiObjectTrackingExample')
% Motion-Based Multiple Object Tracking> and uses the |multiObjectTracker|
% available in Automated Driving System Toolbox.
%
% Copyright 2016 The MathWorks, Inc.
 
%%
% Detection of moving objects and motion-based tracking are important 
% components of many computer vision applications, including activity
% recognition, traffic monitoring, and automotive safety. The problem of
% motion-based object tracking can be divided into two parts:
%
% # Detecting moving objects in each frame 
% # Tracking the moving objects from frame to frame 
%
% The detection of moving objects uses a background subtraction algorithm
% based on Gaussian mixture models. Morphological operations are applied to
% the resulting foreground mask to eliminate noise. Finally, blob analysis
% detects groups of connected pixels, which are likely to correspond to
% moving objects. 
%
% The tracking of moving objects from frame to frame is done by the
% |multiObjectTracker| object that is responsible for the following:
%
% # Assigning detections to tracks. 
% # Initializing new tracks based on unassigned detections. All tracks are
% initialized as |'Tentative'|, accounting for the possibility that they
% resulted from a false detection.
% # Confirming tracks if they have more than _M_ assigned detections in _N_
% frames.
% # Updating existing tracks based on assigned detections.
% # Coasting (predicting) existing unassigned tracks.
% # Deleting tracks if they have remained unassigned (coasted) for too long.
%
% The assignment of detections to the same object is based solely on
% motion. The motion of each track is estimated by a Kalman filter. The
% filter predicts the track's location in each frame, and determines the
% likelihood of each detection being assigned to each track. To initialize
% the filter that you design, use the |FilterInitializationFcn| property of
% the |multiObjectTracker|.
%
% For more information, see
% <matlab:helpview(fullfile(docroot,'toolbox','vision','vision.map'),'multipleObjectTracking') Multiple Object Tracking>.
%
% This example is a function, with the main body at the top and helper 
% routines in the form of 
% <matlab:helpview(fullfile(docroot,'toolbox','matlab','matlab_prog','matlab_prog.map'),'nested_functions') nested functions> 
% below.
 
function p12_on_video_using_tracking_matlab_sample()
% Create objects used for reading video and displaying the results.
videoObjects = setupVideoObjects('6.mp4');
 
% Create objects used for detecting objects in the foreground of the video.
minBlobArea = 10000; % Minimum blob size, in pixels, to be considered as a detection
detectorObjects = setupDetectorObjects(minBlobArea);
 
%% Create the Multi-Object Tracker
% When creating a |multiObjectTracker|, consider the following: 
%
% # |FilterInitializationFcn|: The likely motion and measurement models. 
% In this case, the objects are expected to have a constant speed motion.
% The |initDemoFilter| function configures a linear Kalman filter to 
% track the motion. See the 'Define a Kalman filter' section for details.
% # |AssignmentThreshold|: How far detections may fall from tracks. 
% The default value for this parameter is 30. If there are detections
% that are not assigned to tracks, but should be, increase this value. If
% there are detections that get assigned to tracks that are too far,
% decrease this value.
% # |NumCoastingUpdates|: How long a track is maintained before deletion.
% In this case, since the video has 30 frames per second, a reasonable
% value is about 0.75 seconds (22 frames).
% # |ConfirmationParameters|: The parameters controlling track confirmation.
% A track is initialized with every unassigned detection. Some of these
% detections might be false, so initially, all tracks are |'Tentative'|. 
% To confirm a track, it has to be detected at least _M_ out of _N_
% frames. The choice of _M_ and _N_ depends on the visibility of the
% objects. This example assumes a visibility of 6 out of 10 frames.
tracker = multiObjectTracker(...
 'FilterInitializationFcn', @initDemoFilter, ...
 'AssignmentThreshold', 30, ...
 'NumCoastingUpdates', 22, ...
 'ConfirmationParameters', [6 10] ...
 );
 
%% Define a Kalman Filter
% When defining a tracking filter for the motion, complete the following
% steps:
%
% *Step 1: Define the motion model and state*
%
% In this example, use a constant velocity model in a 2-D rectangular
% frame.
%
% # The state is |[x;vx;y;vy]|.
% # The state transition model matrix is |A = [1 dt 0 0; 0 1 0 0; 0 0 1 dt; 0 0 0 1]|.
% # Assume that |dt = 1|.
%
% *Step 2: Define the process noise*
%
% The process noise represents the parts of the process that are not taken
% into account in the model. For example, in a constant velocity model, the
% acceleration is neglected.
%
% *Step 3: Define the measurement model*
%
% In this example, only the position (|[x;y]|) is measured. So, the
% measurement model is |H = [1 0 0 0; 0 0 1 0]|.
%
% Note: To preconfigure these parameters, define the |'MotionModel'|
% property as |'2D Constant Velocity'|.
%
% *Step 4: Initialize the state vector based on the sensor measurement*
%
% In this example, because the measurement is |[x;y]| and the state is
% |[x;vx;y;vy]|, initializing the state vector is straightforward. Because
% there is no measurement of the velocity, initialize the |vx| and |vy|
% components to 0.
%
% *Step 5: Define an initial state covariance*
%
% In this example, the measurements are quite noisy, so define the initial 
% state covariance to be quite large: |stateCov = diag([50, 50, 50, 50])|
%
% *Step 6: Create the correct filter*
% 
% In this example, all the models are linear, so use |trackingKF| as the
% tracking filter.
 function filter = initDemoFilter(detection)
 % Initialize a Kalman filter for this example.
 
 % Define the initial state.
 state = [detection.Measurement(1); 0; detection.Measurement(2); 0];
 
 % Define the initial state covariance.
 stateCov = diag([50, 50, 50, 50]);
 
 % Create the tracking filter.
 filter = trackingKF('MotionModel', '2D Constant Velocity', ... 
 'State', state, ...
 'StateCovariance', stateCov, ... 
 'MeasurementNoise', detection.MeasurementNoise(1:2,1:2) ... 
 );
 end
 
%%% 
% The following loop runs the video clip, detects moving objects in the
% video, and tracks them across video frames. 
 
% Count frames to create a sense of time.
frameCount = 0;
while hasFrame(videoObjects.reader)
 % Read a video frame and detect objects in it.
 frameCount = frameCount + 1; % Promote frame count
 frame = readFrame(videoObjects.reader); % Read frame 
 [detections, mask] = detectObjects(detectorObjects, frame); % Detect objects in video frame 
 
 % Run the tracker on the preprocessed detections.
 confirmedTracks = updateTracks(tracker, detections, frameCount);
 
 % Display the tracking results on the video.
 displayTrackingResults(videoObjects, confirmedTracks, frame, mask);
end
%% Create Video Objects
% Create objects used for reading and displaying the video frames.
 
 function videoObjects = setupVideoObjects(filename)
 % Initialize video I/O
 % Create objects for reading a video from a file, drawing the tracked
 % objects in each frame, and playing the video.
 
 % Create a video file reader.
 videoObjects.reader = VideoReader(filename);
 
 % Create two video players: one to display the video,
 % and one to display the foreground mask. 
 videoObjects.maskPlayer = vision.VideoPlayer('Position', [20, 400, 700, 400]);
 videoObjects.videoPlayer = vision.VideoPlayer('Position', [740, 400, 700, 400]);
 end
 
%% Create Detector Objects
% Create objects used for detecting foreground objects.
% Use |minBlobArea| to define the size of the blob, in pixels, that is
% considered to be a detection. 
%
% * Increase |minBlobArea| to avoid detecting small blobs, which are more
% likely to be false detections, or if several detections are created for 
% the same object due to partial occlusion.
% * Decrease |minBlobArea| if objects are detected too late or not at all.
 
 function detectorObjects = setupDetectorObjects(minBlobArea)
 % Create System objects for foreground detection and blob analysis
 
 % The foreground detector segments moving objects from the
 % background. It outputs a binary mask, where the pixel value of 1
 % corresponds to the foreground and the value of 0 corresponds to
 % the background.
 
 detectorObjects.detector = vision.ForegroundDetector('NumGaussians', 3, ...
 'NumTrainingFrames', 40, 'MinimumBackgroundRatio', 0.7);
 
 % Connected groups of foreground pixels are likely to correspond to
 % moving objects. The blob analysis System object finds such
 % groups (called 'blobs' or 'connected components') and computes
 % their characteristics, such as their areas, centroids, and the
 % bounding boxes.
 
 detectorObjects.blobAnalyzer = vision.BlobAnalysis('BoundingBoxOutputPort', true, ...
 'AreaOutputPort', true, 'CentroidOutputPort', true, ...
 'MinimumBlobArea', minBlobArea);
 end
 
%% Detect Objects
% The |detectObjects| function returns the centroids and the bounding boxes
% of the detected objects as a list of |objectDetection| objects. You can
% supply this list as an input to the |multiObjectTracker|. The
% |detectObjects| function also returns the binary mask, which has the same
% size as the input frame. Pixels with a value of 1 correspond to the
% foreground. Pixels with a value of 0 correspond to the background.
%
% The function performs motion segmentation using the foreground detector. 
% It then performs morphological operations on the resulting binary mask to
% remove noisy pixels and to fill the holes in the remaining blobs.
%
% When creating the |objectDetection| list, the |frameCount| serves as the
% time input, and the centroids of the detected blobs serve as the
% measurement. The list also has two optional name-value pairs:
%
% * |MeasurementNoise| - Blob detection is noisy, and this example defines 
% a large measurement noise value.
% * |ObjectAttributes| - The detected bounding boxes that get passed to the
% track display are added to this argument.
 
 function [detections, mask] = detectObjects(detectorObjects, frame)
 % Expected uncertainty (noise) for the blob centroid.
 measurementNoise = 100*eye(2); 
 % Detect foreground.
 mask = detectorObjects.detector.step(frame);
 
 % Apply morphological operations to remove noise and fill in holes.
 mask = imopen(mask, strel('rectangle', [9, 9]));
 mask = imclose(mask, strel('rectangle', [10, 10])); 
 mask=bwareaopen(mask,1500);
 mask = imfill(mask, 'holes');
 
 % Perform blob analysis to find connected components.
 [~, centroids, bboxes] = detectorObjects.blobAnalyzer.step(mask);
 
 % Formulate the detections as a list of objectDetection objects.
 numDetections = size(centroids, 1);
 detections = cell(numDetections, 1);
 for i = 1:numDetections
 detections{i} = objectDetection(frameCount, centroids(i,:), ...
 'MeasurementNoise', measurementNoise, ...
 'ObjectAttributes', {bboxes(i,:)});
 end
 end
 
%% Display Tracking Results
% The |displayTrackingResults| function draws a bounding box and label ID
% for each track on the video frame and foreground mask. It then displays
% the frame and the mask in their respective video players.
 
 function displayTrackingResults(videoObjects, confirmedTracks, frame, mask)
 % Convert the frame and the mask to uint8 RGB.
 frame = im2uint8(frame);
 mask = uint8(repmat(mask, [1, 1, 3])) .* 255;
 
 if ~isempty(confirmedTracks) 
 % Display the objects. If an object has not been detected
 % in this frame, display its predicted bounding box.
 numRelTr = numel(confirmedTracks);
 boxes = zeros(numRelTr, 4);
 ids = zeros(numRelTr, 1, 'int32');
 predictedTrackInds = zeros(numRelTr, 1);
 for tr = 1:numRelTr
 % Get bounding boxes.
 boxes(tr, : ) = confirmedTracks(tr).ObjectAttributes{1}{1};
 
 % Get IDs.
 ids(tr) = confirmedTracks(tr).TrackID;
 
 if confirmedTracks(tr).IsCoasted
 predictedTrackInds(tr) = tr;
 end
 end
 
 predictedTrackInds = predictedTrackInds(predictedTrackInds > 0);
 
 % Create labels for objects that display the predicted rather 
 % than the actual location.
 labels = cellstr(int2str(ids));
 
 isPredicted = cell(size(labels));
 isPredicted(predictedTrackInds) = {' predicted'};
 labels = strcat(labels, isPredicted);
 
 % Draw the objects on the frame.
 frame = insertObjectAnnotation(frame, 'rectangle', boxes, labels);
 
 % Draw the objects on the mask.
 mask = insertObjectAnnotation(mask, 'rectangle', boxes, labels);
 end
 
 % Display the mask and the frame.
 videoObjects.maskPlayer.step(mask); 
 videoObjects.videoPlayer.step(frame);
 end
displayEndOfDemoMessage(mfilename)
end
%% Summary
% In this example, you created a motion-based system for detecting and
% tracking multiple moving objects. Try using a different video to see if
% you can detect and track objects. Try modifying the parameters of the
% |multiObjectTracker|.
%
% The tracking in this example was based solely on motion, with the
% assumption that all objects move in a straight line with constant speed.
% When the motion of an object significantly deviates from this model, the
% example can produce tracking errors. Notice the mistake in tracking the
% person occluded by the tree.
%
% You can reduce the likelihood of tracking errors by using a more complex
% motion model, such as constant acceleration or constant turn. To do that,
% try defining a different tracking filter, such as |trackingEKF| or
% |trackingUKF|. 

تشخیص و شمارش خودروهای در حال حرکت

 

منبع:

https://www.mathworks.com/

 

جهت دانلود بر روی لینک زیر کلیک نمایید.

تشخیص و شمارش خودروهای درحال حرکت

ویدئوی پیوست برنامه تشخیص و شمارش خودورهای در حال حرکت

رمز فایل : behsanandish.com

 

محاسبات نرم

محاسبات نرم (به انگلیسی: soft computing) به مجموعه‌ای از شیوه‌های جدید محاسباتی در علوم رایانه، هوش مصنوعی، یادگیری ماشینی و بسیاری از زمینه‌های کاربردی دیگر اطلاق می‌شود. در تمامی این زمینه‌ها به مطالعه، مدل‌سازی و آنالیز پدیده‌های بسیار پیچیده‌ای نیاز است که شیوه‌های علمی دقیق در گذشته، در حل آسان، تحلیلی، و کامل آنها موفق نبوده‌اند.

نکته‌ها و چرایی‌های فلسفی

در مقایسه با تدابیر علمی نرم، روش‌های علمی به‌کاررفته در سده‌های پیشین، تنها از عهدهٔ مدل‌سازی و آنالیز سامانه‌های نسبتاً ساده در مکانیک، فیزیک، و برخی از زمینه‌های کاربردی و مهندسیبرآمده‌اند. مسائل پیچیده‌تری همچون سامانه‌های وابسته به علوم زیست‌شناسی و پزشکی، علوم اجتماعی، علوم انسانی، علوم مدیریت و نظایر آنها بیرون از قلمرو اصلی و توفیق‌آفرین روش‌هایریاضی و تحلیلی دقیق باقی مانده‌بودند. شایان ذکر است که خصایص سادگی و پیچیدگی اموری هستند نسبی، و به‌طور یقین، اغلب مدل‌سازی‌های ریاضی و علمی موفق در گذشته هم، به مفهوم مطلق کلام، بسیار پراهمیت و پیچیده بوده‌اند.

محاسبات نرم با تقبل نادقیق بودن و با محور قراردادن ذهن انسان به‌پیش می‌رود. اصل هدایت‌کنندهٔ محاسبات نرم بهره‌برداری از خاصیت عدم دقیق‌بودن جهت مهارکردن مسئله و پایین‌آوردن هزینهٔ راه‌حل است.

محاسبات نرم را می‌شود حاصل تلاش‌های جدید علمی دانست که مدل‌سازی، تحلیل، و در نهایت کنترل سیستم‌های پیچیده را با سهولت و موفقیت زیادتری امکان‌پذیر می‌سازد. به عنوان مهم‌ترین شاخه‌های این محاسبات، باید منطق فازی، شبکه‌های عصبی مصنوعی، و الگوریتم ژنتیک را بر شمرد.

برخلاف شیوه‌های محاسباتی سخت که تمامی همت و توان خود را به دقیق‌بودن، و “در جهتِ مدل نمودنِ کاملِ حقیقت”، معطوف می‌دارند، روش‌های نرم، براساس تحمل نادقیق‌نگری‌ها، حقایق جزئی و ناکامل، و فقدان اطمینان، استوار گردیده‌اند. درک هرچه روشن‌تر از چرایی، چگونگی، و نیز فلسفهٔ این‌گونه محاسباتِ جدید است که افق‌های جدید در علوم پیچیدهٔ آینده را روشن می‌سازد.

یکی از بزرگ‌ترین زمینه‌های کاربرد محاسبات نرم در ایجاد و گسترش وب معنی‌گرا خواهد بود.

محاسبات نرم در مقایسه با محاسبات سخت به زبان سادهٔ علمی، روش‌های سخت، برآمده از طبیعت و نحوهٔ رفتار ماشین است؛ ولی، در مقابل، شیوه‌های نرم، به انسان و تدابیر اتخاذشده از سویذهن او به‌ منظور حل‌ و فصلِ مسائل، اختصاص پیدا می‌کند.

منبع

 


تعریف محاسبات نرم

شناسایی و نحوه کنترل رفتار یک پدیده و سیستم، از مباحث مهم و کلیدی در امر سیستم کنترل می باشد. اصولاً جهت شناسایی و مدل سازی رفتار یک سیستم به معادله ریاضی آن رجوع می شود. بسیاری از پدیده ها رفتار پیچیده ای دارند و براحتی نمی توان معادله ریاضی آن را بدست آورد. مثلاً نحوه کنترل نوسان بار جرثقیل هوایی جهت قرار دادن بار در نقطه مطلوب، بسیار پیچیده است و اغلب به ۱۰۰% دقیق نیز نخواهد بود و حداقل نیازمند یک معادله دیفرانسل درجه ۵ جهت پیاده سازی آن خواهیم بود.

در صورتی که فقط یک متغیر دیگر بخواهیم به سیستم فوق اضافه نماییم، ممکن است این معادله دیفرانسیل پیچیده تر نیز بشود. بدست آوردن خود این معادله ریاضی دردسر فراوانی دارد، پیاده سازی آن در یک سیستم کنترل الکترونیکی چالش بزرگتری است.

مغز انسان، هرگز برای برخورد با چالش ها و مسائل روزمره پیرامونش، خود را درگیر فرمول و محاسبات پیچیده نمی کند. انسان براساس یادگیری ها و آموخته های خود تصمیماتی می گیرد که منجر به کنترل مسائل پیچیده خواهد بود. اما همین انسان در صورتی که بخواهد نحوه برخورد با یک فرد متخلف را مدل نماید و آن را به یک کامپیوتر بسپارد، حداقل نیاز به دانش معادلات دیفرانسیل، معادلات لاپلاس و … می باشد. اینگونه محاسبات خشک و مبتنی بر معادلات دقیق ریاضیات تحت عنوان “محاسبات سخت” نامیده می شوند.

از طرف دیگر روش هایی وجود دارند که می توانند رفتار پیچیده ترین و مغشوش ترین پدیده ها را نیز با دقت بالایی (نه بصورت ۱۰۰ %دقیق) مدل سازی نمایند. این محاسبات که تحت عنوان “محاسبات نرم” شناخته می شوند، مبتنی بر استنتاج ذهن انسان، شبیه سازی عملکرد نرون های مغز، شبیه سازی رفتار پدیده های اجتماعی طبیعت (الگوریتم های تکاملی مثل ژنتیک، فاخته، کلونی مورچه و…) است.

شبکه های عصبی مصنوعی، سیستم های فازی و الگوریتم های تکاملی از مهمترین شاخه های محاسبات نرم محسوب می شوند. یک راننده ماهر جرثقیل هوایی جهت کنترل نوسان بار، هرگز در ذهنش یک معادله درجه ۵ را بکار نمی گیرد. او با استفاده از یک سیستم استنتاج فازی ذهنی (تعدادی اگر-آنگاه) به خوبی این سیستم پیچیده را با استفاده از تجربیاتش کنترل می نماید. در واقع می توان این تجربیات فرد متخصص را بصورت قوانین فازی درآورد و به سیستم کنترل سپرد.

در سال های اخیر، کاربرد محاسبات نرم در هوش مصنوعی، داده کاوی و  سیستم های کنترل هوشمند بسیار پر رنگ و چاره ساز بوده است.

منابع

  1. https://fa.wikipedia.org
  2. http://mohammadisite.ir

مقالات

1.بررسی روشهای متعادل سازی هیستوگرام در بهبود تصویر

چکیده: افزایش کنتراست به عنوان یکی از مسائل مهم در پردازش تصویر است.متعادل سازی هیستوگرام (HE) یکی از روش های معمول برای بهبود کنتراست در تصاویر دیجیتال است و یک روش افزایش کنتراست ساده و موثر است با این حال، این روش معمولا باعث کنتراست بیش از حد می شود که باعث ظاهر غیر طبیعی در تصویر پردازش شده می شود. هم چنین HE میانگین روشنایی تصویر را به خوبی حفظ نمی کند بنابراین روش های دیگری برای متعادل سازی تصویر با حفظ روشنایی تصویر ارائه شده است. این مقاله به بررسی فرم های جدید هیستوگرام برای افزایش کنتراست تصویر می پردازد. تفاوت عمده میان روش ها معیارهای مورد استفاده، تقسیم هیستوگرام ورودی است. متعادل سازی دو هیستوگرام با حفظ روشنایی(BBHE) میانگین مقادیر شدت به عنوان نقطه جداسازی استفاده می کند. متعادل سازی دو هیستوگرام با حداقل خطای متوسط روشنایی(MMBEBHE) است. متعادل سازی هیستوگرام متوسط – مجرای بازگشتی(RMSHE) بهبود یافته BBHE است. روش یکنواخت سازی پویلی هیستوگرام با حفظ روشنایی(BPDHE) بسط یافته MPHBP و DHE است.
واژه های کلیدی: بهبود کنتراست، متعادل سازی هیستوگرام، خطای متوسط روشنایی، تقسیم بندی هیستوگرام، حفظ روشنایی
فایل PDF – در 22 صفحه- نویسنده : نوشین الله بخشی

بررسی روشهای متعادل سازی هیستوگرام در بهبود تصویر

رمز فایل : behsanandish.com


2. بهبود کیفیت تصاویر آندوسکوپی از طریق تعدیل هیستوگرام فازی و توزیع ناهمسانگرد کنتراست

چکیده: در این مقاله روشی جدید برای بهبود کیفیت تصاویر آندوسکوپی به وسیله ی تعدیل هیستوگرام فازی و توزیع ناهمسانگرد کنتراست ارائه می شود. تصاویر آندوسکوپی موجود در کشورمان از لحاظ نور و کیفیت وضعیت مناسبی ندارند و همین موضوع تبدیل به چالشی جهت تشخیص انواع بیماری های دستگاه گوارش شده است. برای غلبه بر این مشکلات و کمک به پزشکان برای تشخیص بهتر، در این مقاله یک روش وفقی با استفاده از تعدیل هیستوگرام فازی و توزیع کنتراست ارائه می شود. همچنین در روش پیشنهادی مفهوم جدیدی از توزیع کنتراست بر اساس آنالیز محلی تصاویر آندوسکوپی معرفی می شود. سپس به وسیله انتخاب وفقی پارامتر هدایت که نقشی مهم در توزیع ایفا می کند، توزیع کنتراست به منظور بهبود کیفیت تصاویر آندوسکوپی به تصویر اعمال می شود و در نهایت بعد از انتقال به سه فضای رنگ YIQ ،XYZ و HSI به کمک روش تعدیل هیستوگرام فازی، تغییرات نامحسوس رنگ نمایان تر می شود. نتایج تجربی نشان می دهد که روش ارئه شده عملکرد قابل توجهی در افزایش قابلیت دیداری تصاویر آندوسکوپی از خود نشان می دهد.

 

واژه های کلیدی: تصاویر آندوسکوپی، توزیع ناهمسانگرد کنتراست، تعدیل هیستوگرام فازی

فایل PDF – در 6 صفحه- نویسندگان : حسین قیصری، میرحسین دزفولیان

بررسی روشهای متعادل سازی هیستوگرام در بهبود تصویر

رمز فایل : behsanandish.com


3. تشخیص زود هنگام پوسیدگی دندان با استفاده از آنالیز هیستوگرام و طیف توان

چکیده: این مقاله به تشخیص پوسیدگی در مراحل اولیه با استفاده از آنالیز هیستوگرام و طیف توان می پردازد. داده های مورد نیاز شامل تصاویر اشعه ایکس دندان های نرمال و پوسیده از هر شخص می باشد که توسط ابزار پردازش سیگنال MATLAB آنالیز می شود. برای هر تصویر، هیستوگرام و طیف توان محاسبه می شود. سپس یک بررسی دقیق انجام می گیرد. نتایج نشان می دهد که هیستوگرام شدت پیکسل برای دندان های نرمال و پوسیده در محدوده های مختلف متمرکز شده است و تفاوت های آشکاری در بخش های طیفی بدست آمده بین دندان های نرمال و پوسیده وجود دارد. طیف توان دندان پوسیده در مقایسه با طیف دندان نرمال دارای بخش های فرکانس بالااست. هم چنین به کارگیری GUI(واسط کاربر گرافیکی) این کار را آسان تر و وابسته به تعامل کاربر می کند. این روش برای دندان پزشکان در تشخیص پوسیدگی در مراحل اولیه بسیار سودمند می باشد.

واژه های کلیدی: پوسیدگی دندان، هیستوگرام، طیف توان، GUI، شدت پیکسل

فایل PDF – در 6 صفحه- نویسندگان : محمد کریمی مریدانی، شبنم قهاری و فاطمه غلامی

تشخیص زود هنگام پوسیدگی دندان با استفاده از آنالیز هیستوگرام و طیف توان

رمز فایل : behsanandish.com


4. بازیابی تصاویر چهره با استفاده از ترکیب هیستوگرام گرادیان و الگوی باینری محلی

چکیده: بازیابی چهره، یک موضوع تحقیقاتی مهم در پردازش تصویر است که هدف آن استخراج تصاویر چهره ای است که مشابه با یک تصویر جستار باشند. در این مقاله روشی برای بازیابی تصاویر چهره با استفاده از ترکیب هیستوگرام گرادیان و الگوی باینری محلی(LBP) پیشنهاد شده است. ترکیب این دو روش مقاومت در مقابل تغییرات موجود در تصاویر چهره را افزایش می دهد و در نتیجه عملکرد سیستم را در بازیابی تصاویر بهبود می بخشد. برای افزایش توانایی سیستم، یک طرح فیدبک ارتباطی مبتنی بر ماشین بردار پشتیبان(SVM) معرفی می کنیم. آزمایش ها بر روی پایگاه داده ی AR و در دو حالت بدون تصاویر با مانع و با تصاویر با مانع اناجم شده است. نتایج آزمایش ها نشان می دهد که روش پیشنادی ما به خوبی می تواند تصاویر چهره را بازیابی کند. در ادامه، روش پیشنهادی خود را با برخی از روش های موفق در توصیف چهره مقایسه کرده ایم. معیار دقت متوسط میانگین(MAP) برای روش پیشنهادی در حالت های اول و دوم آزمایش به ترتیب برابر است با 94/40%  و 68/12%. در حالی که بهترین نرخ برای روش های مقایسه شده برابر است با 90/37% و 61/99%. این نتایج نشان می دهد روش پیشنهادی ما نسبت به این روش ها بهتر عمل می کند و یک روش خوب برای بازیابی تصاویر چهره است.

واژه های کلیدی: الگوی باینری محلی، بازیابی چهره، فیدبک ارتباطی، ماشین بردار پشتیبان، هیستوگرام گرادیان.

فایل PDF – در 11 صفحه- نویسندگان : محمد قاصری و حسین ابراهیم نژاد

ﺑﺎزﯾﺎﺑﯽ ﺗﺼﺎوﯾﺮ ﭼﻬﺮه ﺑﺎ اﺳﺘﻔﺎده از ﺗﺮﮐﯿﺐ ﻫﯿﺴﺘﻮﮔﺮام ﮔﺮادﯾﺎن و اﻟﮕﻮی ﺑﺎﯾﻨﺮی ﻣﺤﻠﯽ

رمز فایل : behsanandish.com


5. بهبود وفقی کنتراست با استفاده از متعادل سازی بهینه هیستوگرام دو بعدی

چکیده: در این مقاله، برای بهبود وفقی کنتراست به ارائه و حل یک مسئله ی بهینه سازی در هیستوگرام دوبعدی پرداخته شده است. برای جلوگیری از بروز اثرات نامطلوب ناشی از دست کاری هیستوگرام تصویر، در بیان ریاضی مسأله در این مقاله همانند روش های مشابه دیگر، از یک سو هیستوگرام بهینه ی خروجی از روی هیستوگرامی دوبعدی که بیشترین شباهت را به هیستوگرام دوبعدی تصویر ورودی و نیز توزیع یکنواخت داشته باشد به دست می آید و از سویی دیگر بر خلاف دیگر روش ها، با وزن دهی وفقی، اطلاعات محلی مناسبی را نیز دراین جستجو در نظر می گیرد. نگاشت مناسب با حل این مسأله ی بهینه سازی به دست آمده و آزمایش های گوناگونی که بر روی تصاویر گوناگون انجام شده است، درستی مدل بهینه سازی را نشان می دهد. به کارگیری الگوریتم پیشنهادی بر روی تصاویر متعدد، در مقایسه با روش مرجع به صورت میانگین به بهبود 75 درصدی و 3 درصدی معیارهای AMBEN  و  DEN  منجر شده است.

واژه های کلیدی: بهبود کنتراست، هیستوگرام دوبعدی، هموارسازی هیستوگرام

فایل PDF – در 10 صفحه- نویسندگان : سحر ایروانی و مهدی ازوجی

ﺑﻬﺒﻮد وﻓﻘﯽ ﮐﻨﺘﺮاﺳﺖ ﺑﺎ اﺳﺘﻔﺎده از ﻣﺘﻌﺎدل ﺳﺎزی ﺑﻬﯿﻨﻪ ی ﻫﯿﺴﺘﻮﮔﺮام دوﺑﻌﺪی

رمز فایل : behsanandish.com


6. A Study for Applications of Histogram in Image Enhancement

مطالعه برای کاربرد هیستوگرام در بهبود تصویر

Abstract- Image Enhancement aims at improving the visual quality of input image for a particular area. The criterion used by enhancement algorithms to enhance the image is; using histogram details of that image. This paper defines the various applications of histograms through which they help in the enhancement process. The paper also represents three basic histogram processing techniques- histogram sliding, histogram stretching, and histogram equalization, and how these techniques help in enhancement process, which factors effect these techniques. We examine subjectively the effect of these processing techniques. Comparative analysis of these techniques is also carried out.

Keywords: Histogram Equalization, Histogram Sliding, Histogram Stretching, Image Enhancement, Visual Quality.

فایل PDF – در 5 صفحه- نویسندگان : Harpreet Kaur, Neelofar Sohi

A Study for Applications of Histogram in Image Enhancement

رمز فایل : behsanandish.com


7. An Adaptive Histogram Equalization Algorithm on the Image Gray Level Mapping

الگوریتم انعکاس هیستوگرام سازگار بر روی نقشه سطح خاکستری تصویر

Abstract
The conventional histogram equalization algorithm is easy causing information loss. The paper presented an adaptive histogram-based algorithm in which the information entropy remains the same. The algorithm introduces parameter ȕ in the gray level mapping formula, and takes the information entropy as the target function to adaptively adjust the spacing of two adjacent gray levels in the new histogram. So it avoids excessive gray pixel merger and excessive bright local areas of the image. Experiments show that the improved algorithm may effectively improve visual effects under the premise of the same information entropy. It is useful in CT image processing.

Keywords: Histogram Equalization; Image Enhancement; Gray Level Mapping; Information Entropy

فایل PDF – در 8 صفحه- نویسندگان : Youlian Zhu, Cheng Huang

An Adaptive Histogram Equalization Algorithm on the Image

رمز فایل : behsanandish.com


8. Contrast Enhancement Algorithm Based on Gap Adjustment for Histogram Equalization

الگوریتم تقویت کنتراست براساس تنظیم گاف برای برابری هیستوگرام

Abstract: Image enhancement methods have been widely used to improve the visual effects of images. Owing to its simplicity and effectiveness histogram equalization (HE) is one of the methods used for enhancing image contrast. However, HE may result in over-enhancement and feature loss problems that lead to unnatural look and loss of details in the processed images. Researchers have proposed various HE-based methods to solve the over-enhancement problem; however, they have largely ignored the feature loss problem. Therefore, a contrast enhancement algorithm based on gap adjustment for histogram equalization (CegaHE) is proposed. It refers to a visual contrast enhancement algorithm based on histogram equalization (VCEA), which generates visually pleasing enhancedimages,andimprovestheenhancementeffectsofVCEA.CegaHEadjuststhegapsbetween two gray values based on the adjustment equation, which takes the properties of human visual perception into consideration, to solve the over-enhancement problem. Besides, it also alleviates the feature loss problem and further enhances the textures in the dark regions of the images to improve the quality of the processed images for human visual perception. Experimental results demonstrate that CegaHE is a reliable method for contrast enhancement and that it significantly outperforms VCEA and other methods.

Keywords: cumulative distribution function (CDF); contrast enhancement; histogram equalization (HE); human visual perception; gap adjustment

فایل PDF – در 18 صفحه- نویسندگان : Chung-Cheng Chiu , and Chih-Chung Ting

Contrast Enhancement Algorithm Based on Gap

رمز فایل : behsanandish.com


9. Enhancement of Images Using Histogram Processing Techniques

بهبود تصاویر با استفاده از تکنیک های پردازش هیستوگرام

Abstract- Image enhancement is a mean as the improvement of an image appearance by increasing dominance of some features or by decreasing ambiguity between different regions of the image. Image enhancement processes consist of a collection of techniques that seek to improve the visual appearance of an image or to convert the image to a form better suited for analysis by a human or machine. Many images such as medical images, remote sensing images, electron microscopy images and even real life photographic pictures, suffer from poor contrast. Therefore it is necessary to enhance the contrast.The purpose of image enhancement methods is to increase image visibility and details. Enhanced image provide clear image to eyes or assist feature extraction processing in computer vision system. Numerous enhancement methods have been proposed but the enhancement efficiency, computational requirements, noise amplification, user intervention, and application suitability are the common factors to be considered when choosing from these different methods for specific image processing application.

Keywords: Enhancement, Histogram processing techniques, PSNR,MSE.

فایل PDF – در 5 صفحه- نویسندگان :Komal Vij , Yaduvir Singh

Enhancement of Images Using Histogram Processing

رمز فایل : behsanandish.com


10. USE OF HISTOGRAM EQUALIZATION IN IMAGE PROCESSING FOR IMAGE ENHANCEMENT

استفاده از تعادل هیستوگرام در پردازش تصویر برای افزایش تصویر

Abstract— Digital Image Processing is a rapidly evolving field with the growing applications in science & engineering. Image Processing holds the possibility of developing an ultimate machine that could perform visual functions of all living beings. The image processing is a visual task, the foremost step is to obtain an image i.e. image acquisition then enhancement and finally to process. In this paper there are details for image enhancement for the purpose of image processing. Image enhancement is basically improving the digital image quality. Image histogram is helpful in image enhancement. The histogram in the context of image processing is the operation by which the occurrences of each intensity value in the image is shown and Histogram equalization is the technique by which the dynamic range of the histogram of an image is increased.

Keywords- Image processing, image enhancement, image histogram, Histogram equalization

فایل PDF – در 5 صفحه- نویسندگان :Sapana S. Bagade , Vijaya K. Shandilya

USE OF HISTOGRAM EQUALIZATION IN IMAGE PROCESSING FOR IMAGE ENHANCEMENT

رمز فایل : behsanandish.com


جزوات آموزشی

1. Computer Vision – Histogram Processing

1. بینایی کامپیوتر- پردازش هیستوگرام

فایل PDF – در 40 صفحه- نویسنده : Dr. S. Das 

Computer Vision -histogram processing

رمز فایل : behsanandish.com


2. Digital Image Processing (CS/ECE 545)  Lecture 2: Histograms and Point Operations (Part 1)

پردازش تصویر دیجیتال(CS/ECE 545)  درس 2: هیستوگرام و عملیات نقطه

فایل PDF – در 56 صفحه- نویسنده : Prof Emmanuel Agu

Digital Image Processing-histograms and point operations

رمز فایل : behsanandish.com


3. Part 3: Image Processing – Digital Images and Intensity Histograms

بخش 3: پردازش تصویر – تصاویر دیجیتال و هیستوگرام های شدت

فایل PDF – در 57 صفحه- نویسنده : Georgy Gimel’farb

Digital Images and Intensity Histograms

رمز فایل : behsanandish.com


4.  Digital Imaging and Multimedia Histograms of Digital Images

تصویربرداری دیجیتالی و هیستوگرام های چند رسانه ای از تصاویر دیجیتال

فایل PDF – در 12 صفحه- نویسنده : Ahmed Elgammal

Digital Imaging and Multimedia

رمز فایل : behsanandish.com

 

مرحله 4: سرکوب لبه های غیر حداکثر

آخرین مرحله، پیدا کردن لبه های ضعیف که موازی با لبه های قوی هستند و از بین بردن آنهاست. این عمل، با بررسی پیکسل های عمود بر یک پیکسل لبه خاص و حذف لبه های غیر حداکثرانجام شده است. کد مورد استفاده بسیار مشابه کد ردیابی لبه است.

 

&lt;/pre&gt;
&lt;pre&gt;#include "stdafx.h"
#include "tripod.h"
#include "tripodDlg.h"

#include "LVServerDefs.h"
#include "math.h"
#include &lt;fstream&gt;
#include &lt;string&gt;
#include &lt;iostream&gt;
#include &lt;stdlib.h&gt;
#include &lt;stdio.h&gt;


#ifdef _DEBUG
#define new DEBUG_NEW
#undef THIS_FILE
static char THIS_FILE[] = __FILE__;
#endif

using namespace std;

/////////////////////////////////////////////////////////////////////////////
// CAboutDlg dialog used for App About

class CAboutDlg : public CDialog
{
public:
	CAboutDlg();

// Dialog Data
	//{{AFX_DATA(CAboutDlg)
	enum { IDD = IDD_ABOUTBOX };
	//}}AFX_DATA

	// ClassWizard generated virtual function overrides
	//{{AFX_VIRTUAL(CAboutDlg)
	protected:
	virtual void DoDataExchange(CDataExchange* pDX);    // DDX/DDV support
	//}}AFX_VIRTUAL

// Implementation
protected:
	//{{AFX_MSG(CAboutDlg)
	//}}AFX_MSG
	DECLARE_MESSAGE_MAP()
};

CAboutDlg::CAboutDlg() : CDialog(CAboutDlg::IDD)
{
	//{{AFX_DATA_INIT(CAboutDlg)
	//}}AFX_DATA_INIT
}

void CAboutDlg::DoDataExchange(CDataExchange* pDX)
{
	CDialog::DoDataExchange(pDX);
	//{{AFX_DATA_MAP(CAboutDlg)
	//}}AFX_DATA_MAP
}

BEGIN_MESSAGE_MAP(CAboutDlg, CDialog)
	//{{AFX_MSG_MAP(CAboutDlg)
		// No message handlers
	//}}AFX_MSG_MAP
END_MESSAGE_MAP()

/////////////////////////////////////////////////////////////////////////////
// CTripodDlg dialog

CTripodDlg::CTripodDlg(CWnd* pParent /*=NULL*/)
	: CDialog(CTripodDlg::IDD, pParent)
{
	//{{AFX_DATA_INIT(CTripodDlg)
		// NOTE: the ClassWizard will add member initialization here
	//}}AFX_DATA_INIT
	// Note that LoadIcon does not require a subsequent DestroyIcon in Win32
	m_hIcon = AfxGetApp()-&gt;LoadIcon(IDR_MAINFRAME);

	//////////////// Set destination BMP to NULL first 
	m_destinationBitmapInfoHeader = NULL;

}

////////////////////// Additional generic functions

static unsigned PixelBytes(int w, int bpp)
{
    return (w * bpp + 7) / 8;
}

static unsigned DibRowSize(int w, int bpp)
{
    return (w * bpp + 31) / 32 * 4;
}

static unsigned DibRowSize(LPBITMAPINFOHEADER pbi)
{
    return DibRowSize(pbi-&gt;biWidth, pbi-&gt;biBitCount);
}

static unsigned DibRowPadding(int w, int bpp)
{
    return DibRowSize(w, bpp) - PixelBytes(w, bpp);
}

static unsigned DibRowPadding(LPBITMAPINFOHEADER pbi)
{
    return DibRowPadding(pbi-&gt;biWidth, pbi-&gt;biBitCount);
}

static unsigned DibImageSize(int w, int h, int bpp)
{
    return h * DibRowSize(w, bpp);
}

static size_t DibSize(int w, int h, int bpp)
{
    return sizeof (BITMAPINFOHEADER) + DibImageSize(w, h, bpp);
}

/////////////////////// end of generic functions


void CTripodDlg::DoDataExchange(CDataExchange* pDX)
{
	CDialog::DoDataExchange(pDX);
	//{{AFX_DATA_MAP(CTripodDlg)
	DDX_Control(pDX, IDC_PROCESSEDVIEW, m_cVideoProcessedView);
	DDX_Control(pDX, IDC_UNPROCESSEDVIEW, m_cVideoUnprocessedView);
	//}}AFX_DATA_MAP
}

BEGIN_MESSAGE_MAP(CTripodDlg, CDialog)
	//{{AFX_MSG_MAP(CTripodDlg)
	ON_WM_SYSCOMMAND()
	ON_WM_PAINT()
	ON_WM_QUERYDRAGICON()
	ON_BN_CLICKED(IDEXIT, OnExit)
	//}}AFX_MSG_MAP
END_MESSAGE_MAP()

/////////////////////////////////////////////////////////////////////////////
// CTripodDlg message handlers

BOOL CTripodDlg::OnInitDialog()
{
	CDialog::OnInitDialog();

	// Add "About..." menu item to system menu.

	// IDM_ABOUTBOX must be in the system command range.
	ASSERT((IDM_ABOUTBOX &amp; 0xFFF0) == IDM_ABOUTBOX);
	ASSERT(IDM_ABOUTBOX &lt; 0xF000);

	CMenu* pSysMenu = GetSystemMenu(FALSE);
	if (pSysMenu != NULL)
	{
		CString strAboutMenu;
		strAboutMenu.LoadString(IDS_ABOUTBOX);
		if (!strAboutMenu.IsEmpty())
		{
			pSysMenu-&gt;AppendMenu(MF_SEPARATOR);
			pSysMenu-&gt;AppendMenu(MF_STRING, IDM_ABOUTBOX, strAboutMenu);
		}
	}

	// Set the icon for this dialog.  The framework does this automatically
	//  when the application's main window is not a dialog
	SetIcon(m_hIcon, TRUE);			// Set big icon
	SetIcon(m_hIcon, FALSE);		// Set small icon
	
	// TODO: Add extra initialization here

	// For Unprocessed view videoportal (top one)
	char sRegUnprocessedView[] = "HKEY_LOCAL_MACHINE\\Software\\UnprocessedView";
	m_cVideoUnprocessedView.PrepareControl("UnprocessedView", sRegUnprocessedView, 0 );	
	m_cVideoUnprocessedView.EnableUIElements(UIELEMENT_STATUSBAR,0,TRUE);
	m_cVideoUnprocessedView.ConnectCamera2();
	m_cVideoUnprocessedView.SetEnablePreview(TRUE);

	// For binary view videoportal (bottom one)
	char sRegProcessedView[] = "HKEY_LOCAL_MACHINE\\Software\\ProcessedView";
	m_cVideoProcessedView.PrepareControl("ProcessedView", sRegProcessedView, 0 );	
	m_cVideoProcessedView.EnableUIElements(UIELEMENT_STATUSBAR,0,TRUE);
	m_cVideoProcessedView.ConnectCamera2();
	m_cVideoProcessedView.SetEnablePreview(TRUE);

	// Initialize the size of binary videoportal
	m_cVideoProcessedView.SetPreviewMaxHeight(240);
	m_cVideoProcessedView.SetPreviewMaxWidth(320);

	// Uncomment if you wish to fix the live videoportal's size
	// m_cVideoUnprocessedView.SetPreviewMaxHeight(240);
	// m_cVideoUnprocessedView.SetPreviewMaxWidth(320);

	// Find the screen coodinates of the binary videoportal
	m_cVideoProcessedView.GetWindowRect(m_rectForProcessedView);
	ScreenToClient(m_rectForProcessedView);
	allocateDib(CSize(320, 240));

	// Start grabbing frame data for Procssed videoportal (bottom one)
	m_cVideoProcessedView.StartVideoHook(0);

	return TRUE;  // return TRUE  unless you set the focus to a control
}

void CTripodDlg::OnSysCommand(UINT nID, LPARAM lParam)
{
	if ((nID &amp; 0xFFF0) == IDM_ABOUTBOX)
	{
		CAboutDlg dlgAbout;
		dlgAbout.DoModal();
	}
	else
	{
		CDialog::OnSysCommand(nID, lParam);
	}
}

// If you add a minimize button to your dialog, you will need the code below
//  to draw the icon.  For MFC applications using the document/view model,
//  this is automatically done for you by the framework.

void CTripodDlg::OnPaint() 
{
	if (IsIconic())
	{
		CPaintDC dc(this); // device context for painting

		SendMessage(WM_ICONERASEBKGND, (WPARAM) dc.GetSafeHdc(), 0);

		// Center icon in client rectangle
		int cxIcon = GetSystemMetrics(SM_CXICON);
		int cyIcon = GetSystemMetrics(SM_CYICON);
		CRect rect;
		GetClientRect(&amp;rect);
		int x = (rect.Width() - cxIcon + 1) / 2;
		int y = (rect.Height() - cyIcon + 1) / 2;

		// Draw the icon
		dc.DrawIcon(x, y, m_hIcon);
	}
	else
	{
		CDialog::OnPaint();
	}
}

// The system calls this to obtain the cursor to display while the user drags
//  the minimized window.
HCURSOR CTripodDlg::OnQueryDragIcon()
{
	return (HCURSOR) m_hIcon;
}

void CTripodDlg::OnExit() 
{
	// TODO: Add your control notification handler code here

	// Kill live view videoportal (top one)
	m_cVideoUnprocessedView.StopVideoHook(0);
    m_cVideoUnprocessedView.DisconnectCamera();	
	
	// Kill binary view videoportal (bottom one)
	m_cVideoProcessedView.StopVideoHook(0);
    m_cVideoProcessedView.DisconnectCamera();	

	// Kill program
	DestroyWindow();	

	

}

BEGIN_EVENTSINK_MAP(CTripodDlg, CDialog)
    //{{AFX_EVENTSINK_MAP(CTripodDlg)
	ON_EVENT(CTripodDlg, IDC_PROCESSEDVIEW, 1 /* PortalNotification */, OnPortalNotificationProcessedview, VTS_I4 VTS_I4 VTS_I4 VTS_I4)
	//}}AFX_EVENTSINK_MAP
END_EVENTSINK_MAP()

void CTripodDlg::OnPortalNotificationProcessedview(long lMsg, long lParam1, long lParam2, long lParam3) 
{
	// TODO: Add your control notification handler code here
	
	// This function is called at the camera's frame rate
    
#define NOTIFICATIONMSG_VIDEOHOOK	10

	// Declare some useful variables
	// QCSDKMFC.pdf (Quickcam MFC documentation) p. 103 explains the variables lParam1, lParam2, lParam3 too 
	
	LPBITMAPINFOHEADER lpBitmapInfoHeader; // Frame's info header contains info like width and height
	LPBYTE lpBitmapPixelData; // This pointer-to-long will point to the start of the frame's pixel data
    unsigned long lTimeStamp; // Time when frame was grabbed

	switch(lMsg) {
		case NOTIFICATIONMSG_VIDEOHOOK:
			{
				lpBitmapInfoHeader = (LPBITMAPINFOHEADER) lParam1; 
				lpBitmapPixelData = (LPBYTE) lParam2;
				lTimeStamp = (unsigned long) lParam3;

				grayScaleTheFrameData(lpBitmapInfoHeader, lpBitmapPixelData);
				doMyImageProcessing(lpBitmapInfoHeader); // Place where you'd add your image processing code
				displayMyResults(lpBitmapInfoHeader);

			}
			break;

		default:
			break;
	}	
}

void CTripodDlg::allocateDib(CSize sz)
{
	// Purpose: allocate information for a device independent bitmap (DIB)
	// Called from OnInitVideo

	if(m_destinationBitmapInfoHeader) {
		free(m_destinationBitmapInfoHeader);
		m_destinationBitmapInfoHeader = NULL;
	}

	if(sz.cx | sz.cy) {
		m_destinationBitmapInfoHeader = (LPBITMAPINFOHEADER)malloc(DibSize(sz.cx, sz.cy, 24));
		ASSERT(m_destinationBitmapInfoHeader);
		m_destinationBitmapInfoHeader-&gt;biSize = sizeof(BITMAPINFOHEADER);
		m_destinationBitmapInfoHeader-&gt;biWidth = sz.cx;
		m_destinationBitmapInfoHeader-&gt;biHeight = sz.cy;
		m_destinationBitmapInfoHeader-&gt;biPlanes = 1;
		m_destinationBitmapInfoHeader-&gt;biBitCount = 24;
		m_destinationBitmapInfoHeader-&gt;biCompression = 0;
		m_destinationBitmapInfoHeader-&gt;biSizeImage = DibImageSize(sz.cx, sz.cy, 24);
		m_destinationBitmapInfoHeader-&gt;biXPelsPerMeter = 0;
		m_destinationBitmapInfoHeader-&gt;biYPelsPerMeter = 0;
		m_destinationBitmapInfoHeader-&gt;biClrImportant = 0;
		m_destinationBitmapInfoHeader-&gt;biClrUsed = 0;
	}
}

void CTripodDlg::displayMyResults(LPBITMAPINFOHEADER lpThisBitmapInfoHeader)
{
	// displayMyResults: Displays results of doMyImageProcessing() in the videoport
	// Notes: StretchDIBits stretches a device-independent bitmap to the appropriate size

	CDC				*pDC;	// Device context to display bitmap data
	
	pDC = GetDC();	
	int nOldMode = SetStretchBltMode(pDC-&gt;GetSafeHdc(),COLORONCOLOR);

	StretchDIBits( 
		pDC-&gt;GetSafeHdc(),
		m_rectForProcessedView.left,				// videoportal left-most coordinate
		m_rectForProcessedView.top,					// videoportal top-most coordinate
		m_rectForProcessedView.Width(),				// videoportal width
		m_rectForProcessedView.Height(),			// videoportal height
		0,											// Row position to display bitmap in videoportal
		0,											// Col position to display bitmap in videoportal
		lpThisBitmapInfoHeader-&gt;biWidth,			// m_destinationBmp's number of columns
		lpThisBitmapInfoHeader-&gt;biHeight,			// m_destinationBmp's number of rows
		m_destinationBmp,							// The bitmap to display; use the one resulting from doMyImageProcessing
		(BITMAPINFO*)m_destinationBitmapInfoHeader, // The bitmap's header info e.g. width, height, number of bits etc
		DIB_RGB_COLORS,								// Use default 24-bit color table
		SRCCOPY										// Just display
	);
 
	SetStretchBltMode(pDC-&gt;GetSafeHdc(),nOldMode);

	ReleaseDC(pDC);

	// Note: 04/24/02 - Added the following:
	// Christopher Wagner cwagner@fas.harvard.edu noticed that memory wasn't being freed

	// Recall OnPortalNotificationProcessedview, which gets called everytime
	// a frame of data arrives, performs 3 steps:
	// (1) grayScaleTheFrameData - which mallocs m_destinationBmp
	// (2) doMyImageProcesing
	// (3) displayMyResults - which we're in now
	// Since we're finished with the memory we malloc'ed for m_destinationBmp
	// we should free it: 
	
	free(m_destinationBmp);

	// End of adds
}

void CTripodDlg::grayScaleTheFrameData(LPBITMAPINFOHEADER lpThisBitmapInfoHeader, LPBYTE lpThisBitmapPixelData)
{

	// grayScaleTheFrameData: Called by CTripodDlg::OnPortalNotificationBinaryview
	// Task: Read current frame pixel data and computes a grayscale version

	unsigned int	W, H;			  // Width and Height of current frame [pixels]
	BYTE            *sourceBmp;		  // Pointer to current frame of data
	unsigned int    row, col;
	unsigned long   i;
	BYTE			grayValue;

	BYTE			redValue;
	BYTE			greenValue;
	BYTE			blueValue;

    W = lpThisBitmapInfoHeader-&gt;biWidth;  // biWidth: number of columns
    H = lpThisBitmapInfoHeader-&gt;biHeight; // biHeight: number of rows

	// Store pixel data in row-column vector format
	// Recall that each pixel requires 3 bytes (red, blue and green bytes)
	// m_destinationBmp is a protected member and declared in binarizeDlg.h

	m_destinationBmp = (BYTE*)malloc(H*3*W*sizeof(BYTE));

	// Point to the current frame's pixel data
	sourceBmp = lpThisBitmapPixelData;

	for (row = 0; row &lt; H; row++) {
		for (col = 0; col &lt; W; col++) {

			// Recall each pixel is composed of 3 bytes
			i = (unsigned long)(row*3*W + 3*col);
        
			// The source pixel has a blue, green andred value:
			blueValue  = *(sourceBmp + i);
			greenValue = *(sourceBmp + i + 1);
			redValue   = *(sourceBmp + i + 2);

			// A standard equation for computing a grayscale value based on RGB values
			grayValue = (BYTE)(0.299*redValue + 0.587*greenValue + 0.114*blueValue);

			// The destination BMP will be a grayscale version of the source BMP
			*(m_destinationBmp + i)     = grayValue;
			*(m_destinationBmp + i + 1) = grayValue;
			*(m_destinationBmp + i + 2) = grayValue;
			
		}
	}
}


void CTripodDlg::doMyImageProcessing(LPBITMAPINFOHEADER lpThisBitmapInfoHeader)
{
	// doMyImageProcessing:  This is where you'd write your own image processing code
	// Task: Read a pixel's grayscale value and process accordingly

	unsigned int	W, H;			// Width and Height of current frame [pixels]
	unsigned int    row, col;		// Pixel's row and col positions
	unsigned long   i;				// Dummy variable for row-column vector
	int	    upperThreshold = 60;	// Gradient strength nessicary to start edge
	int		lowerThreshold = 30;	// Minimum gradient strength to continue edge
	unsigned long iOffset;			// Variable to offset row-column vector during sobel mask
	int rowOffset;					// Row offset from the current pixel
	int colOffset;					// Col offset from the current pixel
	int rowTotal = 0;				// Row position of offset pixel
	int colTotal = 0;				// Col position of offset pixel
	int Gx;							// Sum of Sobel mask products values in the x direction
	int Gy;							// Sum of Sobel mask products values in the y direction
	float thisAngle;				// Gradient direction based on Gx and Gy
	int newAngle;					// Approximation of the gradient direction
	bool edgeEnd;					// Stores whether or not the edge is at the edge of the possible image
	int GxMask[3][3];				// Sobel mask in the x direction
	int GyMask[3][3];				// Sobel mask in the y direction
	int newPixel;					// Sum pixel values for gaussian
	int gaussianMask[5][5];			// Gaussian mask

	W = lpThisBitmapInfoHeader-&gt;biWidth;  // biWidth: number of columns
    H = lpThisBitmapInfoHeader-&gt;biHeight; // biHeight: number of rows
	
	for (row = 0; row &lt; H; row++) {
		for (col = 0; col &lt; W; col++) {
			edgeDir[row][col] = 0;
		}
	}

	/* Declare Sobel masks */
	GxMask[0][0] = -1; GxMask[0][1] = 0; GxMask[0][2] = 1;
	GxMask[1][0] = -2; GxMask[1][1] = 0; GxMask[1][2] = 2;
	GxMask[2][0] = -1; GxMask[2][1] = 0; GxMask[2][2] = 1;
	
	GyMask[0][0] =  1; GyMask[0][1] =  2; GyMask[0][2] =  1;
	GyMask[1][0] =  0; GyMask[1][1] =  0; GyMask[1][2] =  0;
	GyMask[2][0] = -1; GyMask[2][1] = -2; GyMask[2][2] = -1;

	/* Declare Gaussian mask */
	gaussianMask[0][0] = 2;		gaussianMask[0][1] = 4;		gaussianMask[0][2] = 5;		gaussianMask[0][3] = 4;		gaussianMask[0][4] = 2;	
	gaussianMask[1][0] = 4;		gaussianMask[1][1] = 9;		gaussianMask[1][2] = 12;	gaussianMask[1][3] = 9;		gaussianMask[1][4] = 4;	
	gaussianMask[2][0] = 5;		gaussianMask[2][1] = 12;	gaussianMask[2][2] = 15;	gaussianMask[2][3] = 12;	gaussianMask[2][4] = 2;	
	gaussianMask[3][0] = 4;		gaussianMask[3][1] = 9;		gaussianMask[3][2] = 12;	gaussianMask[3][3] = 9;		gaussianMask[3][4] = 4;	
	gaussianMask[4][0] = 2;		gaussianMask[4][1] = 4;		gaussianMask[4][2] = 5;		gaussianMask[4][3] = 4;		gaussianMask[4][4] = 2;	
	

	/* Gaussian Blur */
	for (row = 2; row &lt; H-2; row++) {
		for (col = 2; col &lt; W-2; col++) {
			newPixel = 0;
			for (rowOffset=-2; rowOffset&lt;=2; rowOffset++) {
				for (colOffset=-2; colOffset&lt;=2; colOffset++) {
					rowTotal = row + rowOffset;
					colTotal = col + colOffset;
					iOffset = (unsigned long)(rowTotal*3*W + colTotal*3);
					newPixel += (*(m_destinationBmp + iOffset)) * gaussianMask[2 + rowOffset][2 + colOffset];
				}
			}
			i = (unsigned long)(row*3*W + col*3);
			*(m_destinationBmp + i) = newPixel / 159;
		}
	}

	/* Determine edge directions and gradient strengths */
	for (row = 1; row &lt; H-1; row++) {
		for (col = 1; col &lt; W-1; col++) {
			i = (unsigned long)(row*3*W + 3*col);
			Gx = 0;
			Gy = 0;
			/* Calculate the sum of the Sobel mask times the nine surrounding pixels in the x and y direction */
			for (rowOffset=-1; rowOffset&lt;=1; rowOffset++) {
				for (colOffset=-1; colOffset&lt;=1; colOffset++) {
					rowTotal = row + rowOffset;
					colTotal = col + colOffset;
					iOffset = (unsigned long)(rowTotal*3*W + colTotal*3);
					Gx = Gx + (*(m_destinationBmp + iOffset) * GxMask[rowOffset + 1][colOffset + 1]);
					Gy = Gy + (*(m_destinationBmp + iOffset) * GyMask[rowOffset + 1][colOffset + 1]);
				}
			}

			gradient[row][col] = sqrt(pow(Gx,2.0) + pow(Gy,2.0));	// Calculate gradient strength			
			thisAngle = (atan2(Gx,Gy)/3.14159) * 180.0;		// Calculate actual direction of edge
			
			/* Convert actual edge direction to approximate value */
			if ( ( (thisAngle &lt; 22.5) &amp;&amp; (thisAngle &gt; -22.5) ) || (thisAngle &gt; 157.5) || (thisAngle &lt; -157.5) )
				newAngle = 0;
			if ( ( (thisAngle &gt; 22.5) &amp;&amp; (thisAngle &lt; 67.5) ) || ( (thisAngle &lt; -112.5) &amp;&amp; (thisAngle &gt; -157.5) ) )
				newAngle = 45;
			if ( ( (thisAngle &gt; 67.5) &amp;&amp; (thisAngle &lt; 112.5) ) || ( (thisAngle &lt; -67.5) &amp;&amp; (thisAngle &gt; -112.5) ) )
				newAngle = 90;
			if ( ( (thisAngle &gt; 112.5) &amp;&amp; (thisAngle &lt; 157.5) ) || ( (thisAngle &lt; -22.5) &amp;&amp; (thisAngle &gt; -67.5) ) )
				newAngle = 135;
				
			edgeDir[row][col] = newAngle;		// Store the approximate edge direction of each pixel in one array
		}
	}

	/* Trace along all the edges in the image */
	for (row = 1; row &lt; H - 1; row++) {
		for (col = 1; col &lt; W - 1; col++) {
			edgeEnd = false;
			if (gradient[row][col] &gt; upperThreshold) {		// Check to see if current pixel has a high enough gradient strength to be part of an edge
				/* Switch based on current pixel's edge direction */
				switch (edgeDir[row][col]){		
					case 0:
						findEdge(0, 1, row, col, 0, lowerThreshold);
						break;
					case 45:
						findEdge(1, 1, row, col, 45, lowerThreshold);
						break;
					case 90:
						findEdge(1, 0, row, col, 90, lowerThreshold);
						break;
					case 135:
						findEdge(1, -1, row, col, 135, lowerThreshold);
						break;
					default :
						i = (unsigned long)(row*3*W + 3*col);
						*(m_destinationBmp + i) = 
						*(m_destinationBmp + i + 1) = 
						*(m_destinationBmp + i + 2) = 0;
						break;
					}
				}
			else {
				i = (unsigned long)(row*3*W + 3*col);
					*(m_destinationBmp + i) = 
					*(m_destinationBmp + i + 1) = 
					*(m_destinationBmp + i + 2) = 0;
			}	
		}
	}
	
	/* Suppress any pixels not changed by the edge tracing */
	for (row = 0; row &lt; H; row++) {
		for (col = 0; col &lt; W; col++) {	
			// Recall each pixel is composed of 3 bytes
			i = (unsigned long)(row*3*W + 3*col);
			// If a pixel's grayValue is not black or white make it black
			if( ((*(m_destinationBmp + i) != 255) &amp;&amp; (*(m_destinationBmp + i) != 0)) || ((*(m_destinationBmp + i + 1) != 255) &amp;&amp; (*(m_destinationBmp + i + 1) != 0)) || ((*(m_destinationBmp + i + 2) != 255) &amp;&amp; (*(m_destinationBmp + i + 2) != 0)) ) 
				*(m_destinationBmp + i) = 
				*(m_destinationBmp + i + 1) = 
				*(m_destinationBmp + i + 2) = 0; // Make pixel black
		}
	}

	/* Non-maximum Suppression */
	for (row = 1; row &lt; H - 1; row++) {
		for (col = 1; col &lt; W - 1; col++) {
			i = (unsigned long)(row*3*W + 3*col);
			if (*(m_destinationBmp + i) == 255) {		// Check to see if current pixel is an edge
				/* Switch based on current pixel's edge direction */
				switch (edgeDir[row][col]) {		
					case 0:
						suppressNonMax( 1, 0, row, col, 0, lowerThreshold);
						break;
					case 45:
						suppressNonMax( 1, -1, row, col, 45, lowerThreshold);
						break;
					case 90:
						suppressNonMax( 0, 1, row, col, 90, lowerThreshold);
						break;
					case 135:
						suppressNonMax( 1, 1, row, col, 135, lowerThreshold);
						break;
					default :
						break;
				}
			}	
		}
	}
	
}

void CTripodDlg::findEdge(int rowShift, int colShift, int row, int col, int dir, int lowerThreshold)
{
	int W = 320;
	int H = 240;
	int newRow;
	int newCol;
	unsigned long i;
	bool edgeEnd = false;

	/* Find the row and column values for the next possible pixel on the edge */
	if (colShift &lt; 0) {
		if (col &gt; 0)
			newCol = col + colShift;
		else
			edgeEnd = true;
	} else if (col &lt; W - 1) {
		newCol = col + colShift;
	} else
		edgeEnd = true;		// If the next pixel would be off image, don't do the while loop
	if (rowShift &lt; 0) {
		if (row &gt; 0)
			newRow = row + rowShift;
		else
			edgeEnd = true;
	} else if (row &lt; H - 1) {
		newRow = row + rowShift;
	} else
		edgeEnd = true;	
		
	/* Determine edge directions and gradient strengths */
	while ( (edgeDir[newRow][newCol]==dir) &amp;&amp; !edgeEnd &amp;&amp; (gradient[newRow][newCol] &gt; lowerThreshold) ) {
		/* Set the new pixel as white to show it is an edge */
		i = (unsigned long)(newRow*3*W + 3*newCol);
		*(m_destinationBmp + i) =
		*(m_destinationBmp + i + 1) =
		*(m_destinationBmp + i + 2) = 255;
		if (colShift &lt; 0) {
			if (newCol &gt; 0)
				newCol = newCol + colShift;
			else
				edgeEnd = true;	
		} else if (newCol &lt; W - 1) {
			newCol = newCol + colShift;
		} else
			edgeEnd = true;	
		if (rowShift &lt; 0) {
			if (newRow &gt; 0)
				newRow = newRow + rowShift;
			else
				edgeEnd = true;
		} else if (newRow &lt; H - 1) {
			newRow = newRow + rowShift;
		} else
			edgeEnd = true;	
	}	
}

void CTripodDlg::suppressNonMax(int rowShift, int colShift, int row, int col, int dir, int lowerThreshold)
{
	int W = 320;
	int H = 240;
	int newRow = 0;
	int newCol = 0;
	unsigned long i;
	bool edgeEnd = false;
	float nonMax[320][3];			// Temporarily stores gradients and positions of pixels in parallel edges
	int pixelCount = 0;					// Stores the number of pixels in parallel edges
	int count;						// A for loop counter
	int max[3];						// Maximum point in a wide edge
	
	if (colShift &lt; 0) {
		if (col &gt; 0)
			newCol = col + colShift;
		else
			edgeEnd = true;
	} else if (col &lt; W - 1) {
		newCol = col + colShift;
	} else
		edgeEnd = true;		// If the next pixel would be off image, don't do the while loop
	if (rowShift &lt; 0) {
		if (row &gt; 0)
			newRow = row + rowShift;
		else
			edgeEnd = true;
	} else if (row &lt; H - 1) {
		newRow = row + rowShift;
	} else
		edgeEnd = true;	
	i = (unsigned long)(newRow*3*W + 3*newCol);
	/* Find non-maximum parallel edges tracing up */
	while ((edgeDir[newRow][newCol] == dir) &amp;&amp; !edgeEnd &amp;&amp; (*(m_destinationBmp + i) == 255)) {
		if (colShift &lt; 0) {
			if (newCol &gt; 0)
				newCol = newCol + colShift;
			else
				edgeEnd = true;	
		} else if (newCol &lt; W - 1) {
			newCol = newCol + colShift;
		} else
			edgeEnd = true;	
		if (rowShift &lt; 0) {
			if (newRow &gt; 0)
				newRow = newRow + rowShift;
			else
				edgeEnd = true;
		} else if (newRow &lt; H - 1) {
			newRow = newRow + rowShift;
		} else
			edgeEnd = true;	
		nonMax[pixelCount][0] = newRow;
		nonMax[pixelCount][1] = newCol;
		nonMax[pixelCount][2] = gradient[newRow][newCol];
		pixelCount++;
		i = (unsigned long)(newRow*3*W + 3*newCol);
	}

	/* Find non-maximum parallel edges tracing down */
	edgeEnd = false;
	colShift *= -1;
	rowShift *= -1;
	if (colShift &lt; 0) {
		if (col &gt; 0)
			newCol = col + colShift;
		else
			edgeEnd = true;
	} else if (col &lt; W - 1) {
		newCol = col + colShift;
	} else
		edgeEnd = true;	
	if (rowShift &lt; 0) {
		if (row &gt; 0)
			newRow = row + rowShift;
		else
			edgeEnd = true;
	} else if (row &lt; H - 1) {
		newRow = row + rowShift;
	} else
		edgeEnd = true;	
	i = (unsigned long)(newRow*3*W + 3*newCol);
	while ((edgeDir[newRow][newCol] == dir) &amp;&amp; !edgeEnd &amp;&amp; (*(m_destinationBmp + i) == 255)) {
		if (colShift &lt; 0) {
			if (newCol &gt; 0)
				newCol = newCol + colShift;
			else
				edgeEnd = true;	
		} else if (newCol &lt; W - 1) {
			newCol = newCol + colShift;
		} else
			edgeEnd = true;	
		if (rowShift &lt; 0) {
			if (newRow &gt; 0)
				newRow = newRow + rowShift;
			else
				edgeEnd = true;
		} else if (newRow &lt; H - 1) {
			newRow = newRow + rowShift;
		} else
			edgeEnd = true;	
		nonMax[pixelCount][0] = newRow;
		nonMax[pixelCount][1] = newCol;
		nonMax[pixelCount][2] = gradient[newRow][newCol];
		pixelCount++;
		i = (unsigned long)(newRow*3*W + 3*newCol);
	}

	/* Suppress non-maximum edges */
	max[0] = 0;
	max[1] = 0;
	max[2] = 0;
	for (count = 0; count &lt; pixelCount; count++) {
		if (nonMax[count][2] &gt; max[2]) {
			max[0] = nonMax[count][0];
			max[1] = nonMax[count][1];
			max[2] = nonMax[count][2];
		}
	}
	for (count = 0; count &lt; pixelCount; count++) {
		i = (unsigned long)(nonMax[count][0]*3*W + 3*nonMax[count][1]);
		*(m_destinationBmp + i) = 
		*(m_destinationBmp + i + 1) = 
		*(m_destinationBmp + i + 2) = 0;
	}
}

 

دانلود کد فوق از طریق لینک زیر:

Canny in C++ -No2

رمز فایل : behsanandish.com

الگوریتم Canny در سی پلاس پلاس قسمت 1
الگوریتم Canny در سی پلاس پلاس قسمت 2
الگوریتم Canny در سی پلاس پلاس قسمت 3
الگوریتم Canny در سی پلاس پلاس قسمت 4

مرحله 2: پیدا کردن قدرت و جهت گرادیان لبه.

گام بعدی استفاده از Mask های Sobel برای پیدا کردن قدرت و جهت گرادیان لبه برای هر پیکسل است. ابتدا ماسک های Sobel به محدوده پیکسل 3×3 پیکسل فعلی در هر دو جهت x و y اعمال می شود. سپس مجموع مقدار هر ماسک ضربدر پیکسل مربوطه به ترتیب به عنوان مقادیر Gx و Gy محاسبه می شود. ریشه دوم مربع Gx به اضافه Gy مربع برابر قدرت لبه است. Tangent معکوس Gx / Gy جهت لبه را تولید می کند. سپس جهت لبه تقریب شده است به یکی از چهار مقادیر ممکن که ایجاد می کند جهت های ممکن را که  یک لبه می تواند در یک تصویر از یک شبکه پیکسل مربع باشد. این جهت لبه در edgeDir [row] [col] ذخیره می شود و قدرت گرادیان در  array gradient[row] [col] ذخیره می شود.

 

CannyEdgeWeel

هر زاویه لبه در 11.25 درجه از یکی از  زاویه های ممکن به آن مقدار تغییر می کند.

 

#include "stdafx.h"
#include "tripod.h"
#include "tripodDlg.h"

#include "LVServerDefs.h"
#include "math.h"
#include &lt;fstream&gt;
#include &lt;string&gt;
#include &lt;iostream&gt;
#include &lt;stdlib.h&gt;
#include &lt;stdio.h&gt;


#ifdef _DEBUG
#define new DEBUG_NEW
#undef THIS_FILE
static char THIS_FILE[] = __FILE__;
#endif

using namespace std;

/////////////////////////////////////////////////////////////////////////////
// CAboutDlg dialog used for App About

class CAboutDlg : public CDialog
{
public:
	CAboutDlg();

// Dialog Data
	//{{AFX_DATA(CAboutDlg)
	enum { IDD = IDD_ABOUTBOX };
	//}}AFX_DATA

	// ClassWizard generated virtual function overrides
	//{{AFX_VIRTUAL(CAboutDlg)
	protected:
	virtual void DoDataExchange(CDataExchange* pDX);    // DDX/DDV support
	//}}AFX_VIRTUAL

// Implementation
protected:
	//{{AFX_MSG(CAboutDlg)
	//}}AFX_MSG
	DECLARE_MESSAGE_MAP()
};

CAboutDlg::CAboutDlg() : CDialog(CAboutDlg::IDD)
{
	//{{AFX_DATA_INIT(CAboutDlg)
	//}}AFX_DATA_INIT
}

void CAboutDlg::DoDataExchange(CDataExchange* pDX)
{
	CDialog::DoDataExchange(pDX);
	//{{AFX_DATA_MAP(CAboutDlg)
	//}}AFX_DATA_MAP
}

BEGIN_MESSAGE_MAP(CAboutDlg, CDialog)
	//{{AFX_MSG_MAP(CAboutDlg)
		// No message handlers
	//}}AFX_MSG_MAP
END_MESSAGE_MAP()

/////////////////////////////////////////////////////////////////////////////
// CTripodDlg dialog

CTripodDlg::CTripodDlg(CWnd* pParent /*=NULL*/)
	: CDialog(CTripodDlg::IDD, pParent)
{
	//{{AFX_DATA_INIT(CTripodDlg)
		// NOTE: the ClassWizard will add member initialization here
	//}}AFX_DATA_INIT
	// Note that LoadIcon does not require a subsequent DestroyIcon in Win32
	m_hIcon = AfxGetApp()-&gt;LoadIcon(IDR_MAINFRAME);

	//////////////// Set destination BMP to NULL first 
	m_destinationBitmapInfoHeader = NULL;

}

////////////////////// Additional generic functions

static unsigned PixelBytes(int w, int bpp)
{
    return (w * bpp + 7) / 8;
}

static unsigned DibRowSize(int w, int bpp)
{
    return (w * bpp + 31) / 32 * 4;
}

static unsigned DibRowSize(LPBITMAPINFOHEADER pbi)
{
    return DibRowSize(pbi-&gt;biWidth, pbi-&gt;biBitCount);
}

static unsigned DibRowPadding(int w, int bpp)
{
    return DibRowSize(w, bpp) - PixelBytes(w, bpp);
}

static unsigned DibRowPadding(LPBITMAPINFOHEADER pbi)
{
    return DibRowPadding(pbi-&gt;biWidth, pbi-&gt;biBitCount);
}

static unsigned DibImageSize(int w, int h, int bpp)
{
    return h * DibRowSize(w, bpp);
}

static size_t DibSize(int w, int h, int bpp)
{
    return sizeof (BITMAPINFOHEADER) + DibImageSize(w, h, bpp);
}

/////////////////////// end of generic functions


void CTripodDlg::DoDataExchange(CDataExchange* pDX)
{
	CDialog::DoDataExchange(pDX);
	//{{AFX_DATA_MAP(CTripodDlg)
	DDX_Control(pDX, IDC_PROCESSEDVIEW, m_cVideoProcessedView);
	DDX_Control(pDX, IDC_UNPROCESSEDVIEW, m_cVideoUnprocessedView);
	//}}AFX_DATA_MAP
}

BEGIN_MESSAGE_MAP(CTripodDlg, CDialog)
	//{{AFX_MSG_MAP(CTripodDlg)
	ON_WM_SYSCOMMAND()
	ON_WM_PAINT()
	ON_WM_QUERYDRAGICON()
	ON_BN_CLICKED(IDEXIT, OnExit)
	//}}AFX_MSG_MAP
END_MESSAGE_MAP()

/////////////////////////////////////////////////////////////////////////////
// CTripodDlg message handlers

BOOL CTripodDlg::OnInitDialog()
{
	CDialog::OnInitDialog();

	// Add "About..." menu item to system menu.

	// IDM_ABOUTBOX must be in the system command range.
	ASSERT((IDM_ABOUTBOX &amp; 0xFFF0) == IDM_ABOUTBOX);
	ASSERT(IDM_ABOUTBOX &lt; 0xF000);

	CMenu* pSysMenu = GetSystemMenu(FALSE);
	if (pSysMenu != NULL)
	{
		CString strAboutMenu;
		strAboutMenu.LoadString(IDS_ABOUTBOX);
		if (!strAboutMenu.IsEmpty())
		{
			pSysMenu-&gt;AppendMenu(MF_SEPARATOR);
			pSysMenu-&gt;AppendMenu(MF_STRING, IDM_ABOUTBOX, strAboutMenu);
		}
	}

	// Set the icon for this dialog.  The framework does this automatically
	//  when the application's main window is not a dialog
	SetIcon(m_hIcon, TRUE);			// Set big icon
	SetIcon(m_hIcon, FALSE);		// Set small icon
	
	// TODO: Add extra initialization here

	// For Unprocessed view videoportal (top one)
	char sRegUnprocessedView[] = "HKEY_LOCAL_MACHINE\\Software\\UnprocessedView";
	m_cVideoUnprocessedView.PrepareControl("UnprocessedView", sRegUnprocessedView, 0 );	
	m_cVideoUnprocessedView.EnableUIElements(UIELEMENT_STATUSBAR,0,TRUE);
	m_cVideoUnprocessedView.ConnectCamera2();
	m_cVideoUnprocessedView.SetEnablePreview(TRUE);

	// For binary view videoportal (bottom one)
	char sRegProcessedView[] = "HKEY_LOCAL_MACHINE\\Software\\ProcessedView";
	m_cVideoProcessedView.PrepareControl("ProcessedView", sRegProcessedView, 0 );	
	m_cVideoProcessedView.EnableUIElements(UIELEMENT_STATUSBAR,0,TRUE);
	m_cVideoProcessedView.ConnectCamera2();
	m_cVideoProcessedView.SetEnablePreview(TRUE);

	// Initialize the size of binary videoportal
	m_cVideoProcessedView.SetPreviewMaxHeight(240);
	m_cVideoProcessedView.SetPreviewMaxWidth(320);

	// Uncomment if you wish to fix the live videoportal's size
	// m_cVideoUnprocessedView.SetPreviewMaxHeight(240);
	// m_cVideoUnprocessedView.SetPreviewMaxWidth(320);

	// Find the screen coodinates of the binary videoportal
	m_cVideoProcessedView.GetWindowRect(m_rectForProcessedView);
	ScreenToClient(m_rectForProcessedView);
	allocateDib(CSize(320, 240));

	// Start grabbing frame data for Procssed videoportal (bottom one)
	m_cVideoProcessedView.StartVideoHook(0);

	return TRUE;  // return TRUE  unless you set the focus to a control
}

void CTripodDlg::OnSysCommand(UINT nID, LPARAM lParam)
{
	if ((nID &amp; 0xFFF0) == IDM_ABOUTBOX)
	{
		CAboutDlg dlgAbout;
		dlgAbout.DoModal();
	}
	else
	{
		CDialog::OnSysCommand(nID, lParam);
	}
}

// If you add a minimize button to your dialog, you will need the code below
//  to draw the icon.  For MFC applications using the document/view model,
//  this is automatically done for you by the framework.

void CTripodDlg::OnPaint() 
{
	if (IsIconic())
	{
		CPaintDC dc(this); // device context for painting

		SendMessage(WM_ICONERASEBKGND, (WPARAM) dc.GetSafeHdc(), 0);

		// Center icon in client rectangle
		int cxIcon = GetSystemMetrics(SM_CXICON);
		int cyIcon = GetSystemMetrics(SM_CYICON);
		CRect rect;
		GetClientRect(&amp;rect);
		int x = (rect.Width() - cxIcon + 1) / 2;
		int y = (rect.Height() - cyIcon + 1) / 2;

		// Draw the icon
		dc.DrawIcon(x, y, m_hIcon);
	}
	else
	{
		CDialog::OnPaint();
	}
}

// The system calls this to obtain the cursor to display while the user drags
//  the minimized window.
HCURSOR CTripodDlg::OnQueryDragIcon()
{
	return (HCURSOR) m_hIcon;
}

void CTripodDlg::OnExit() 
{
	// TODO: Add your control notification handler code here

	// Kill live view videoportal (top one)
	m_cVideoUnprocessedView.StopVideoHook(0);
    m_cVideoUnprocessedView.DisconnectCamera();	
	
	// Kill binary view videoportal (bottom one)
	m_cVideoProcessedView.StopVideoHook(0);
    m_cVideoProcessedView.DisconnectCamera();	

	// Kill program
	DestroyWindow();	

	

}

BEGIN_EVENTSINK_MAP(CTripodDlg, CDialog)
    //{{AFX_EVENTSINK_MAP(CTripodDlg)
	ON_EVENT(CTripodDlg, IDC_PROCESSEDVIEW, 1 /* PortalNotification */, OnPortalNotificationProcessedview, VTS_I4 VTS_I4 VTS_I4 VTS_I4)
	//}}AFX_EVENTSINK_MAP
END_EVENTSINK_MAP()

void CTripodDlg::OnPortalNotificationProcessedview(long lMsg, long lParam1, long lParam2, long lParam3) 
{
	// TODO: Add your control notification handler code here
	
	// This function is called at the camera's frame rate
    
#define NOTIFICATIONMSG_VIDEOHOOK	10

	// Declare some useful variables
	// QCSDKMFC.pdf (Quickcam MFC documentation) p. 103 explains the variables lParam1, lParam2, lParam3 too 
	
	LPBITMAPINFOHEADER lpBitmapInfoHeader; // Frame's info header contains info like width and height
	LPBYTE lpBitmapPixelData; // This pointer-to-long will point to the start of the frame's pixel data
    unsigned long lTimeStamp; // Time when frame was grabbed

	switch(lMsg) {
		case NOTIFICATIONMSG_VIDEOHOOK:
			{
				lpBitmapInfoHeader = (LPBITMAPINFOHEADER) lParam1; 
				lpBitmapPixelData = (LPBYTE) lParam2;
				lTimeStamp = (unsigned long) lParam3;

				grayScaleTheFrameData(lpBitmapInfoHeader, lpBitmapPixelData);
				doMyImageProcessing(lpBitmapInfoHeader); // Place where you'd add your image processing code
				displayMyResults(lpBitmapInfoHeader);

			}
			break;

		default:
			break;
	}	
}

void CTripodDlg::allocateDib(CSize sz)
{
	// Purpose: allocate information for a device independent bitmap (DIB)
	// Called from OnInitVideo

	if(m_destinationBitmapInfoHeader) {
		free(m_destinationBitmapInfoHeader);
		m_destinationBitmapInfoHeader = NULL;
	}

	if(sz.cx | sz.cy) {
		m_destinationBitmapInfoHeader = (LPBITMAPINFOHEADER)malloc(DibSize(sz.cx, sz.cy, 24));
		ASSERT(m_destinationBitmapInfoHeader);
		m_destinationBitmapInfoHeader-&gt;biSize = sizeof(BITMAPINFOHEADER);
		m_destinationBitmapInfoHeader-&gt;biWidth = sz.cx;
		m_destinationBitmapInfoHeader-&gt;biHeight = sz.cy;
		m_destinationBitmapInfoHeader-&gt;biPlanes = 1;
		m_destinationBitmapInfoHeader-&gt;biBitCount = 24;
		m_destinationBitmapInfoHeader-&gt;biCompression = 0;
		m_destinationBitmapInfoHeader-&gt;biSizeImage = DibImageSize(sz.cx, sz.cy, 24);
		m_destinationBitmapInfoHeader-&gt;biXPelsPerMeter = 0;
		m_destinationBitmapInfoHeader-&gt;biYPelsPerMeter = 0;
		m_destinationBitmapInfoHeader-&gt;biClrImportant = 0;
		m_destinationBitmapInfoHeader-&gt;biClrUsed = 0;
	}
}

void CTripodDlg::displayMyResults(LPBITMAPINFOHEADER lpThisBitmapInfoHeader)
{
	// displayMyResults: Displays results of doMyImageProcessing() in the videoport
	// Notes: StretchDIBits stretches a device-independent bitmap to the appropriate size

	CDC				*pDC;	// Device context to display bitmap data
	
	pDC = GetDC();	
	int nOldMode = SetStretchBltMode(pDC-&gt;GetSafeHdc(),COLORONCOLOR);

	StretchDIBits( 
		pDC-&gt;GetSafeHdc(),
		m_rectForProcessedView.left,				// videoportal left-most coordinate
		m_rectForProcessedView.top,					// videoportal top-most coordinate
		m_rectForProcessedView.Width(),				// videoportal width
		m_rectForProcessedView.Height(),			// videoportal height
		0,											// Row position to display bitmap in videoportal
		0,											// Col position to display bitmap in videoportal
		lpThisBitmapInfoHeader-&gt;biWidth,			// m_destinationBmp's number of columns
		lpThisBitmapInfoHeader-&gt;biHeight,			// m_destinationBmp's number of rows
		m_destinationBmp,							// The bitmap to display; use the one resulting from doMyImageProcessing
		(BITMAPINFO*)m_destinationBitmapInfoHeader, // The bitmap's header info e.g. width, height, number of bits etc
		DIB_RGB_COLORS,								// Use default 24-bit color table
		SRCCOPY										// Just display
	);
 
	SetStretchBltMode(pDC-&gt;GetSafeHdc(),nOldMode);

	ReleaseDC(pDC);

	// Note: 04/24/02 - Added the following:
	// Christopher Wagner cwagner@fas.harvard.edu noticed that memory wasn't being freed

	// Recall OnPortalNotificationProcessedview, which gets called everytime
	// a frame of data arrives, performs 3 steps:
	// (1) grayScaleTheFrameData - which mallocs m_destinationBmp
	// (2) doMyImageProcesing
	// (3) displayMyResults - which we're in now
	// Since we're finished with the memory we malloc'ed for m_destinationBmp
	// we should free it: 
	
	free(m_destinationBmp);

	// End of adds
}

void CTripodDlg::grayScaleTheFrameData(LPBITMAPINFOHEADER lpThisBitmapInfoHeader, LPBYTE lpThisBitmapPixelData)
{

	// grayScaleTheFrameData: Called by CTripodDlg::OnPortalNotificationBinaryview
	// Task: Read current frame pixel data and computes a grayscale version

	unsigned int	W, H;			  // Width and Height of current frame [pixels]
	BYTE            *sourceBmp;		  // Pointer to current frame of data
	unsigned int    row, col;
	unsigned long   i;
	BYTE			grayValue;

	BYTE			redValue;
	BYTE			greenValue;
	BYTE			blueValue;

    W = lpThisBitmapInfoHeader-&gt;biWidth;  // biWidth: number of columns
    H = lpThisBitmapInfoHeader-&gt;biHeight; // biHeight: number of rows

	// Store pixel data in row-column vector format
	// Recall that each pixel requires 3 bytes (red, blue and green bytes)
	// m_destinationBmp is a protected member and declared in binarizeDlg.h

	m_destinationBmp = (BYTE*)malloc(H*3*W*sizeof(BYTE));

	// Point to the current frame's pixel data
	sourceBmp = lpThisBitmapPixelData;

	for (row = 0; row &lt; H; row++) {
		for (col = 0; col &lt; W; col++) {

			// Recall each pixel is composed of 3 bytes
			i = (unsigned long)(row*3*W + 3*col);
        
			// The source pixel has a blue, green andred value:
			blueValue  = *(sourceBmp + i);
			greenValue = *(sourceBmp + i + 1);
			redValue   = *(sourceBmp + i + 2);

			// A standard equation for computing a grayscale value based on RGB values
			grayValue = (BYTE)(0.299*redValue + 0.587*greenValue + 0.114*blueValue);

			// The destination BMP will be a grayscale version of the source BMP
			*(m_destinationBmp + i)     = grayValue;
			*(m_destinationBmp + i + 1) = grayValue;
			*(m_destinationBmp + i + 2) = grayValue;
			
		}
	}
}


void CTripodDlg::doMyImageProcessing(LPBITMAPINFOHEADER lpThisBitmapInfoHeader)
{
	// doMyImageProcessing:  This is where you'd write your own image processing code
	// Task: Read a pixel's grayscale value and process accordingly

	unsigned int	W, H;			// Width and Height of current frame [pixels]
	unsigned int    row, col;		// Pixel's row and col positions
	unsigned long   i;				// Dummy variable for row-column vector
	int	    upperThreshold = 60;	// Gradient strength nessicary to start edge
	int		lowerThreshold = 30;	// Minimum gradient strength to continue edge
	unsigned long iOffset;			// Variable to offset row-column vector during sobel mask
	int rowOffset;					// Row offset from the current pixel
	int colOffset;					// Col offset from the current pixel
	int rowTotal = 0;				// Row position of offset pixel
	int colTotal = 0;				// Col position of offset pixel
	int Gx;							// Sum of Sobel mask products values in the x direction
	int Gy;							// Sum of Sobel mask products values in the y direction
	float thisAngle;				// Gradient direction based on Gx and Gy
	int newAngle;					// Approximation of the gradient direction
	bool edgeEnd;					// Stores whether or not the edge is at the edge of the possible image
	int GxMask[3][3];				// Sobel mask in the x direction
	int GyMask[3][3];				// Sobel mask in the y direction
	int newPixel;					// Sum pixel values for gaussian
	int gaussianMask[5][5];			// Gaussian mask

	W = lpThisBitmapInfoHeader-&gt;biWidth;  // biWidth: number of columns
    H = lpThisBitmapInfoHeader-&gt;biHeight; // biHeight: number of rows
	
	for (row = 0; row &lt; H; row++) {
		for (col = 0; col &lt; W; col++) {
			edgeDir[row][col] = 0;
		}
	}

	/* Declare Sobel masks */
	GxMask[0][0] = -1; GxMask[0][1] = 0; GxMask[0][2] = 1;
	GxMask[1][0] = -2; GxMask[1][1] = 0; GxMask[1][2] = 2;
	GxMask[2][0] = -1; GxMask[2][1] = 0; GxMask[2][2] = 1;
	
	GyMask[0][0] =  1; GyMask[0][1] =  2; GyMask[0][2] =  1;
	GyMask[1][0] =  0; GyMask[1][1] =  0; GyMask[1][2] =  0;
	GyMask[2][0] = -1; GyMask[2][1] = -2; GyMask[2][2] = -1;

	/* Declare Gaussian mask */
	gaussianMask[0][0] = 2;		gaussianMask[0][1] = 4;		gaussianMask[0][2] = 5;		gaussianMask[0][3] = 4;		gaussianMask[0][4] = 2;	
	gaussianMask[1][0] = 4;		gaussianMask[1][1] = 9;		gaussianMask[1][2] = 12;	gaussianMask[1][3] = 9;		gaussianMask[1][4] = 4;	
	gaussianMask[2][0] = 5;		gaussianMask[2][1] = 12;	gaussianMask[2][2] = 15;	gaussianMask[2][3] = 12;	gaussianMask[2][4] = 2;	
	gaussianMask[3][0] = 4;		gaussianMask[3][1] = 9;		gaussianMask[3][2] = 12;	gaussianMask[3][3] = 9;		gaussianMask[3][4] = 4;	
	gaussianMask[4][0] = 2;		gaussianMask[4][1] = 4;		gaussianMask[4][2] = 5;		gaussianMask[4][3] = 4;		gaussianMask[4][4] = 2;	
	

	/* Gaussian Blur */
	for (row = 2; row &lt; H-2; row++) {
		for (col = 2; col &lt; W-2; col++) {
			newPixel = 0;
			for (rowOffset=-2; rowOffset&lt;=2; rowOffset++) {
				for (colOffset=-2; colOffset&lt;=2; colOffset++) {
					rowTotal = row + rowOffset;
					colTotal = col + colOffset;
					iOffset = (unsigned long)(rowTotal*3*W + colTotal*3);
					newPixel += (*(m_destinationBmp + iOffset)) * gaussianMask[2 + rowOffset][2 + colOffset];
				}
			}
			i = (unsigned long)(row*3*W + col*3);
			*(m_destinationBmp + i) = newPixel / 159;
		}
	}

	/* Determine edge directions and gradient strengths */
	for (row = 1; row &lt; H-1; row++) {
		for (col = 1; col &lt; W-1; col++) {
			i = (unsigned long)(row*3*W + 3*col);
			Gx = 0;
			Gy = 0;
			/* Calculate the sum of the Sobel mask times the nine surrounding pixels in the x and y direction */
			for (rowOffset=-1; rowOffset&lt;=1; rowOffset++) {
				for (colOffset=-1; colOffset&lt;=1; colOffset++) {
					rowTotal = row + rowOffset;
					colTotal = col + colOffset;
					iOffset = (unsigned long)(rowTotal*3*W + colTotal*3);
					Gx = Gx + (*(m_destinationBmp + iOffset) * GxMask[rowOffset + 1][colOffset + 1]);
					Gy = Gy + (*(m_destinationBmp + iOffset) * GyMask[rowOffset + 1][colOffset + 1]);
				}
			}

			gradient[row][col] = sqrt(pow(Gx,2.0) + pow(Gy,2.0));	// Calculate gradient strength			
			thisAngle = (atan2(Gx,Gy)/3.14159) * 180.0;		// Calculate actual direction of edge
			
			/* Convert actual edge direction to approximate value */
			if ( ( (thisAngle &lt; 22.5) &amp;&amp; (thisAngle &gt; -22.5) ) || (thisAngle &gt; 157.5) || (thisAngle &lt; -157.5) )
				newAngle = 0;
			if ( ( (thisAngle &gt; 22.5) &amp;&amp; (thisAngle &lt; 67.5) ) || ( (thisAngle &lt; -112.5) &amp;&amp; (thisAngle &gt; -157.5) ) )
				newAngle = 45;
			if ( ( (thisAngle &gt; 67.5) &amp;&amp; (thisAngle &lt; 112.5) ) || ( (thisAngle &lt; -67.5) &amp;&amp; (thisAngle &gt; -112.5) ) )
				newAngle = 90;
			if ( ( (thisAngle &gt; 112.5) &amp;&amp; (thisAngle &lt; 157.5) ) || ( (thisAngle &lt; -22.5) &amp;&amp; (thisAngle &gt; -67.5) ) )
				newAngle = 135;
				
			edgeDir[row][col] = newAngle;		// Store the approximate edge direction of each pixel in one array
		}
	}

	/* Trace along all the edges in the image */
	for (row = 1; row &lt; H - 1; row++) {
		for (col = 1; col &lt; W - 1; col++) {
			edgeEnd = false;
			if (gradient[row][col] &gt; upperThreshold) {		// Check to see if current pixel has a high enough gradient strength to be part of an edge
				/* Switch based on current pixel's edge direction */
				switch (edgeDir[row][col]){		
					case 0:
						findEdge(0, 1, row, col, 0, lowerThreshold);
						break;
					case 45:
						findEdge(1, 1, row, col, 45, lowerThreshold);
						break;
					case 90:
						findEdge(1, 0, row, col, 90, lowerThreshold);
						break;
					case 135:
						findEdge(1, -1, row, col, 135, lowerThreshold);
						break;
					default :
						i = (unsigned long)(row*3*W + 3*col);
						*(m_destinationBmp + i) = 
						*(m_destinationBmp + i + 1) = 
						*(m_destinationBmp + i + 2) = 0;
						break;
					}
				}
			else {
				i = (unsigned long)(row*3*W + 3*col);
					*(m_destinationBmp + i) = 
					*(m_destinationBmp + i + 1) = 
					*(m_destinationBmp + i + 2) = 0;
			}	
		}
	}
	
	/* Suppress any pixels not changed by the edge tracing */
	for (row = 0; row &lt; H; row++) {
		for (col = 0; col &lt; W; col++) {	
			// Recall each pixel is composed of 3 bytes
			i = (unsigned long)(row*3*W + 3*col);
			// If a pixel's grayValue is not black or white make it black
			if( ((*(m_destinationBmp + i) != 255) &amp;&amp; (*(m_destinationBmp + i) != 0)) || ((*(m_destinationBmp + i + 1) != 255) &amp;&amp; (*(m_destinationBmp + i + 1) != 0)) || ((*(m_destinationBmp + i + 2) != 255) &amp;&amp; (*(m_destinationBmp + i + 2) != 0)) ) 
				*(m_destinationBmp + i) = 
				*(m_destinationBmp + i + 1) = 
				*(m_destinationBmp + i + 2) = 0; // Make pixel black
		}
	}

	/* Non-maximum Suppression */
	for (row = 1; row &lt; H - 1; row++) {
		for (col = 1; col &lt; W - 1; col++) {
			i = (unsigned long)(row*3*W + 3*col);
			if (*(m_destinationBmp + i) == 255) {		// Check to see if current pixel is an edge
				/* Switch based on current pixel's edge direction */
				switch (edgeDir[row][col]) {		
					case 0:
						suppressNonMax( 1, 0, row, col, 0, lowerThreshold);
						break;
					case 45:
						suppressNonMax( 1, -1, row, col, 45, lowerThreshold);
						break;
					case 90:
						suppressNonMax( 0, 1, row, col, 90, lowerThreshold);
						break;
					case 135:
						suppressNonMax( 1, 1, row, col, 135, lowerThreshold);
						break;
					default :
						break;
				}
			}	
		}
	}
	
}

void CTripodDlg::findEdge(int rowShift, int colShift, int row, int col, int dir, int lowerThreshold)
{
	int W = 320;
	int H = 240;
	int newRow;
	int newCol;
	unsigned long i;
	bool edgeEnd = false;

	/* Find the row and column values for the next possible pixel on the edge */
	if (colShift &lt; 0) {
		if (col &gt; 0)
			newCol = col + colShift;
		else
			edgeEnd = true;
	} else if (col &lt; W - 1) {
		newCol = col + colShift;
	} else
		edgeEnd = true;		// If the next pixel would be off image, don't do the while loop
	if (rowShift &lt; 0) {
		if (row &gt; 0)
			newRow = row + rowShift;
		else
			edgeEnd = true;
	} else if (row &lt; H - 1) {
		newRow = row + rowShift;
	} else
		edgeEnd = true;	
		
	/* Determine edge directions and gradient strengths */
	while ( (edgeDir[newRow][newCol]==dir) &amp;&amp; !edgeEnd &amp;&amp; (gradient[newRow][newCol] &gt; lowerThreshold) ) {
		/* Set the new pixel as white to show it is an edge */
		i = (unsigned long)(newRow*3*W + 3*newCol);
		*(m_destinationBmp + i) =
		*(m_destinationBmp + i + 1) =
		*(m_destinationBmp + i + 2) = 255;
		if (colShift &lt; 0) {
			if (newCol &gt; 0)
				newCol = newCol + colShift;
			else
				edgeEnd = true;	
		} else if (newCol &lt; W - 1) {
			newCol = newCol + colShift;
		} else
			edgeEnd = true;	
		if (rowShift &lt; 0) {
			if (newRow &gt; 0)
				newRow = newRow + rowShift;
			else
				edgeEnd = true;
		} else if (newRow &lt; H - 1) {
			newRow = newRow + rowShift;
		} else
			edgeEnd = true;	
	}	
}

void CTripodDlg::suppressNonMax(int rowShift, int colShift, int row, int col, int dir, int lowerThreshold)
{
	int W = 320;
	int H = 240;
	int newRow = 0;
	int newCol = 0;
	unsigned long i;
	bool edgeEnd = false;
	float nonMax[320][3];			// Temporarily stores gradients and positions of pixels in parallel edges
	int pixelCount = 0;					// Stores the number of pixels in parallel edges
	int count;						// A for loop counter
	int max[3];						// Maximum point in a wide edge
	
	if (colShift &lt; 0) {
		if (col &gt; 0)
			newCol = col + colShift;
		else
			edgeEnd = true;
	} else if (col &lt; W - 1) {
		newCol = col + colShift;
	} else
		edgeEnd = true;		// If the next pixel would be off image, don't do the while loop
	if (rowShift &lt; 0) {
		if (row &gt; 0)
			newRow = row + rowShift;
		else
			edgeEnd = true;
	} else if (row &lt; H - 1) {
		newRow = row + rowShift;
	} else
		edgeEnd = true;	
	i = (unsigned long)(newRow*3*W + 3*newCol);
	/* Find non-maximum parallel edges tracing up */
	while ((edgeDir[newRow][newCol] == dir) &amp;&amp; !edgeEnd &amp;&amp; (*(m_destinationBmp + i) == 255)) {
		if (colShift &lt; 0) {
			if (newCol &gt; 0)
				newCol = newCol + colShift;
			else
				edgeEnd = true;	
		} else if (newCol &lt; W - 1) {
			newCol = newCol + colShift;
		} else
			edgeEnd = true;	
		if (rowShift &lt; 0) {
			if (newRow &gt; 0)
				newRow = newRow + rowShift;
			else
				edgeEnd = true;
		} else if (newRow &lt; H - 1) {
			newRow = newRow + rowShift;
		} else
			edgeEnd = true;	
		nonMax[pixelCount][0] = newRow;
		nonMax[pixelCount][1] = newCol;
		nonMax[pixelCount][2] = gradient[newRow][newCol];
		pixelCount++;
		i = (unsigned long)(newRow*3*W + 3*newCol);
	}

	/* Find non-maximum parallel edges tracing down */
	edgeEnd = false;
	colShift *= -1;
	rowShift *= -1;
	if (colShift &lt; 0) {
		if (col &gt; 0)
			newCol = col + colShift;
		else
			edgeEnd = true;
	} else if (col &lt; W - 1) {
		newCol = col + colShift;
	} else
		edgeEnd = true;	
	if (rowShift &lt; 0) {
		if (row &gt; 0)
			newRow = row + rowShift;
		else
			edgeEnd = true;
	} else if (row &lt; H - 1) {
		newRow = row + rowShift;
	} else
		edgeEnd = true;	
	i = (unsigned long)(newRow*3*W + 3*newCol);
	while ((edgeDir[newRow][newCol] == dir) &amp;&amp; !edgeEnd &amp;&amp; (*(m_destinationBmp + i) == 255)) {
		if (colShift &lt; 0) {
			if (newCol &gt; 0)
				newCol = newCol + colShift;
			else
				edgeEnd = true;	
		} else if (newCol &lt; W - 1) {
			newCol = newCol + colShift;
		} else
			edgeEnd = true;	
		if (rowShift &lt; 0) {
			if (newRow &gt; 0)
				newRow = newRow + rowShift;
			else
				edgeEnd = true;
		} else if (newRow &lt; H - 1) {
			newRow = newRow + rowShift;
		} else
			edgeEnd = true;	
		nonMax[pixelCount][0] = newRow;
		nonMax[pixelCount][1] = newCol;
		nonMax[pixelCount][2] = gradient[newRow][newCol];
		pixelCount++;
		i = (unsigned long)(newRow*3*W + 3*newCol);
	}

	/* Suppress non-maximum edges */
	max[0] = 0;
	max[1] = 0;
	max[2] = 0;
	for (count = 0; count &lt; pixelCount; count++) {
		if (nonMax[count][2] &gt; max[2]) {
			max[0] = nonMax[count][0];
			max[1] = nonMax[count][1];
			max[2] = nonMax[count][2];
		}
	}
	for (count = 0; count &lt; pixelCount; count++) {
		i = (unsigned long)(nonMax[count][0]*3*W + 3*nonMax[count][1]);
		*(m_destinationBmp + i) = 
		*(m_destinationBmp + i + 1) = 
		*(m_destinationBmp + i + 2) = 0;
	}
}

الگوریتم Canny در سی پلاس پلاس قسمت 1
الگوریتم Canny در سی پلاس پلاس قسمت 2
الگوریتم Canny در سی پلاس پلاس قسمت 3
الگوریتم Canny در سی پلاس پلاس قسمت 4