different stages of using OpenCV

1: read in images.

Mat src = imread("and.png");

2: convert to a gray image.

 Mat gray;

cvtColor(src, gray, CV_BGR2GRAY);

3: find edges. There are two ways: threshold and canny. This step is optional. If the picture is only black and white, then this step is not necessary. We can just use findContours.

Mat th; Mat bw;

//method 1:  threshold(gray, th,125, 255,THRESH_BINARY);

//method 2:  Canny(gray, bw, 100, 200); 

       Note the differences between the above two images.
4: find contours and draw contours.

4.1:  findContours(th, contours, hierarchy,CV_RETR_TREE, CV_CHAIN_APPROX_SIMPLE);

 Note the red rectangle. If we change :

threshold(gray, th,125, 255,THRESH_BINARY);

to

threshold(gray, th,125, 255,THRESH_BINARY_INV);

You will not see that red  rectangle.

   4.2: findContours(bw, contours2, hierarchy2,CV_RETR_TREE, CV_CHAIN_APPROX_SIMPLE);

Why do we get 4 contours instead of 2 contours since we only two shapes in the canny output?
Because Canny creates two contours for each edge, one outside and one inside of that edge.


4: write out the image.

imwrite("output.png", color);