First we look at the range of the delta E values:
max(errorImage(:)),For this example, most of the errors are less than 10, as this histogram shows (below). Only a small number of pixels have delta E values greater than 20:
min(errorImage(:))
hist(errorImage(:),[1:2:14]),
sum(errorImage(:) > 20), % We think this is 173
Now let's study the spatial distribution of the errors and just mark the ones that are 10 or larger using green (above):
errorTruncated = min(128*(errorImage/10),128*ones(size(errorImage)));If you have the image processing toolbox, you can find out where the edges are and overlay the edges with the locations of the scielab errors. The edges are marked in red in the image above.
figure;
colormap([gray(127); [0 1 0]]);
colormap([gray(127); [1 1 1]]);
image(errorTruncated);
edgeImage = 129 * double(edge(rHats,'prewitt'));return to the S-CIELAB main page
comparison = max(edgeImage,errorTruncated);
mp = [gray(127); [0 1 0]; [1 0 0] ];
colormap(mp);
image(comparison);