I am using cv.jit.centroids to find objects in a quicktime movie. the movie is reduced to binary black and white, and then small red crosses are superimposed upon the centers of any objects in the movie with green circles centered on the dots representing area. I want just the green and the red, without the binary black and white part, so I subtracted the binary window from the window with both the binary and the color overlays, in theory giving me a movie with just the green and red parts left. The problem is that the remainder is broken up. not all the pixels are there, so the circles are broken and the crosses incomplete. Is there a better way or a fix for this? Here is an image of the subpatch:
if i get it right you just need red and green lcd image. If that is the case you need to disconnect right inlet from cv.jit.centroids.draw but be careful because you gave dimension information through that inlet. Break it after you sand the image.