Project (OneDrive): https://onedrive.live.com/embed?cid=DEDCB6EF190B8FD4&resid=DEDCB6EF190B8FD4%21411&authkey=ANtICXVGKPsQ030
Introduction
Several times, I have read questions regarding how to use OpenCV in MFC projects, so I decided to made a simple application demo, a VS2010 project, that reveals a way to use computer vision library inside of VC/MFC project, developed in an MDI application. Through my findings, as far as I remember, I saw somewhere a CDialog based app that embeds OpenCV ... here is another approach: multi-document interface, where you can load multiple media files.
Background
This article does not treat the OpenCV compilation, it does contain the compiled OpenCV library, ready to go, inside of two folders: include, and res86, all of them are included inside the project. On the internet, you can easily find the way in which to compile OpenCV, this is not the purpose of this article.
Step to Setup the Project
1. Put "input" and "x86" folders inside your project:
2. Setup project settings as follows:
Additional include directories: C:\Flaviu\Tempo\include;%(AdditionalIncludeDirectories) (replace "C:\Flaviu\" with your path).
Additional library directories: C:\Flaviu\Tempo\x86\vc10\lib;%(AdditionalLibraryDirectories) (replace "C:\Flaviu\" with your path).
Additional dependencies:
opencv_calib3d341d.lib
opencv_core341d.lib
opencv_features2d341d.lib
opencv_flann341d.lib
opencv_highgui341d.lib
opencv_imgcodecs341d.lib
opencv_imgproc341d.lib
opencv_ml341d.lib
opencv_objdetect341d.lib
opencv_photo341d.lib
opencv_shape341d.lib
opencv_stitching341d.lib
opencv_superres341d.lib
opencv_video341d.lib
opencv_videoio341d.lib
opencv_videostab341d.lib
in Debug case, and the same libs for Release but without "d" from file names. Example: opencv_calib3d341.lib as opencv_calib3d341d.lib.
Using the Code
The structure of the project is pretty simple: the document class contains a cv::VideoCapture object, used to read video files, a cv::Mat object that is needed to convert cv::VideoCapture into cv::Mat object, and a BITMAPINFO structure used to convert cv::Mat object into bitmap object in order to be used for CView rendering. This conversion could be found on CTempoDoc::SetupBitmapInfo method.
void CTempoDoc::SetupBitmapInfo(cv::Mat& mat)
{
if(NULL != m_pBmi)
{
delete m_pBmi;
m_pBmi = NULL;
}
m_pBmi = new BITMAPINFO;
BITMAPINFOHEADER* pHeader = &m_pBmi->bmiHeader;
pHeader->biSize = sizeof(BITMAPINFOHEADER);
pHeader->biPlanes = 1;
pHeader->biCompression = BI_RGB;
pHeader->biXPelsPerMeter = 100;
pHeader->biYPelsPerMeter = 100;
pHeader->biClrUsed = 0;
pHeader->biClrImportant = 0;
pHeader->biWidth = m_Mat.cols;
pHeader->biHeight = -m_Mat.rows;
pHeader->biBitCount = 24;
m_pBmi->bmiHeader.biSizeImage = 0;
}
I should say that CTempoDoc has a tricky method: ResizeMat(cv::Mat& mat) ... this method corrects the cv::Mat that does not have the right format (actually, add a new column to matrix OpenCV object retrieved from cv:imread, when it is needed).
The rendering is handled inside CView class. Here, you can spot the well known OnDraw method, which renders the content of CTempoDoc::BITMAPINFO content, double-buffered. And a timer is implemented for the case when the loaded file is a video stream, and this timer will know when to load and render the next frame from video. Also, this class contains another handler that takes care of how the content is erased and drawn on CView, when the size of the window has been changed or when the next frame of video is ready to render, and some methods that scale the images inside window. All of them can be explored inside the sample project.
Here, you have an application that can load static images or all kinds of videos into an MDI app. Enjoy it!
Points of Interest
The computer vision library is used more and more, and I offer a way here to include your OpenCV algorithms inside a real life desktop application.
History
- 11th September, 2018 - Published the article
