Interactive Rectangle Selection

Demonstrate using a mouse to interact with an image. Allows the user to select parts of an image with a mouse. When they let go of the mouse, it correlates (using cv.matchTemplate) that patch with the image.

The sample uses the RectSelector class.

Sources:

https://github.com/opencv/opencv/blob/3.3.1/samples/python/mouse_and_match.py

load and show image

fname = fullfile(mexopencv.root(), 'test', 'fruits.jpg');
img = cv.imread(fname, 'Color',true);
gray = cv.cvtColor(img, 'RGB2GRAY');
hImg = imshow(gray);

prompt user to select ROI rectangle

disp('Select part of the image using the mouse');
if true
    roi = RectSelector(hImg);
    roi.clip = true;
    setLineProps(roi, 'Color','g', 'LineWidth',2);
    rect = wait(roi);
    delete(roi);
else
    roi = imrect(ancestor(hImg,'axes'));
    setColor(roi, 'g');
    rect = wait(roi);
    delete(roi);
end
assert(~isempty(rect), 'No selection');
rectangle('Position',rect, 'EdgeColor','g', 'LineWidth',2)

Select part of the image using the mouse

extract patch and perform template matching against image

tmpl = cv.Rect.crop(gray, rect);
res = cv.matchTemplate(gray, tmpl, 'Method','CCoeffNormed');
res = abs(res).^3;
res = cv.threshold(res, 0.01, 'Type','ToZero');
res8 = cv.normalize(res, 'Alpha',0, 'Beta',255, ...
    'NormType','MinMax', 'DType','uint8');
figure, imshow(res8)