I want to compare Screenshots of different versions (same site) of a web-application. Until now I tried following approaches:
- pixel based comparison
- pixel based fuzzy comparison with Delta E
- segmented based comparison
For the latter I subdivided the screenshot in different parts of the same size and compared them via color histograms.
Now I want to subdivide the image in a more intelligent way. Maybe I could divide it in some regions of interest. Can anyone explain some approaches for that please?
Here is an example for my problem: