Question

I am doing a project on COmputer Vision and I have a question. After I have successfully loaded a video using OpenCV, I would like to extract the video frames and view their similarities. I have been told I need to use a matrix to achieve this purpose. How would I go about doing this?

Answer 1

When you capture a video frame you get an instance of the Mat Class. This is already a Matrix of your video frames images pixels.

To view similarities you will need to capture a frame and then capture another, once you have two frames you can run a comparison.

You will need to explain further exactly what you want to achieve as there are lots of ways to compare images and get different results using two frames of a video capture.

Mat videoFrame, previousFrame;
VideoSource myCam;
// initialise your video frame.
prepVideoCam();

// start a loop{
    myCam >> videoFrame; // Captures the current frame to videoFrame.
    if(i_have_previousFrame)
        // do some processing if i have a previous frame.

    videoFrame.copyTo(previousFrame);
// back to loop

You will find that videoFrame is a matrix already.

To help you on your way start here http://code.opencv.org/projects/opencv/wiki if you haven’t already done so. OpenCV moved from an old website to a new one and i believe this is the latest.

Download this document http://docs.opencv.org/opencv_tutorials.pdf

This document will explain the differences and provide you with lots of tutorials to work your way through.

The section “2.1 Mat – The Basic Image Container” begins to explain the differences and updates to openCV between v1 and v2. It also shows how to use them interchangeably, but personally stick to v2 syntax and classes.

Take your time in this section as it can quickly get complicated.

Also note that when you install opencv, all of the source files for these tutorials i think are a part of the package, or at least there is a file somewhere with them all in.

Hope that helps further.