Stereo vision: Coresponding coordinates

Question

Please could you help me understand to the code from Matlab Help about Stereo vision - Basic Block Matching?

% Scan over all rows.
for m=1:nRowsLeft
    % Set min/max row bounds for image block.
    minr = max(1,m-halfBlockSize);
    maxr = min(nRowsLeft,m+halfBlockSize);
    % Scan over all columns.
    for n=1:size(leftI,2)
        minc = max(1,n-halfBlockSize);
        maxc = min(size(leftI,2),n+halfBlockSize);
        % Compute disparity bounds.
        mind = max( -disparityRange, 1-minc );
        maxd = min( disparityRange, size(leftI,2)-maxc );

        % Construct template and region of interest.
        template = rightI(minr:maxr,minc:maxc);
        templateCenter = floor((size(template)+1)/2);
        roi = [minc+templateCenter(2)+mind-1 ...
               minr+templateCenter(1)-1 ...
               maxd-mind+1 1];
        % Lookup proper TemplateMatcher object; create if empty.
        if isempty(tmats{size(template,1),size(template,2)})
            tmats{size(template,1),size(template,2)} = ...
                vision.TemplateMatcher('ROIInputPort',true);
        end
        thisTemplateMatcher = tmats{size(template,1),size(template,2)};

        % Run TemplateMatcher object.
        loc = step(thisTemplateMatcher, leftI, template, roi);
        Dbasic(m,n) = loc(1) - roi(1) + mind;
    end
    waitbar(m/nRowsLeft,hWaitBar);
end

According to Matlab help: loc contains location coordinates of the point in the left image and roi contains [x y width height] of the region in the right image where x and y are coordinates of the upper left corner.

I want to know corresponding coordinates in the left and right image. For the left image: x = loc(1), y = loc(2) and for the right image: x = roi(1) - mind, y = loc(2)

Is this correct? I am not sure what exactly is inside roi.

Answer 1

You should be able to use the disparity function from the Computer Vision System Toolbox instead, which implements the block-matching algorithm.

If you just need the matching points, you should use local features: detect some interest points in both images, like corners or blobs, extract features descriptors from patches around those points, and then match up the descriptors. This is called finding correspondences, and the advantage of this method is that your images do not need to be rectified.

Here is an example of finding correspondences to estimate an affine transformation between a pair of images. Here is an example of finding correspondences for stereo rectification.