Quantifying a picture difference?
For image verification, our application under test needs to be compared with several different baseline images due to factors that can't be controlled by transparency, masking, color tolerance, or pixel tolerance.
It would be convenient if we were able to quantify which of the different baselines was the closest match if none are close enough to be acceptable.
Are there any suggestions for determining this in an efficient manner? I can see there would be some terrible ways to do it, like iteratively increasing the pixel tolerance until picture.compare passes or summing together the RGB values of every pixel in a picture.difference result - which aren't really options.
It would be convenient if we were able to quantify which of the different baselines was the closest match if none are close enough to be acceptable.
Are there any suggestions for determining this in an efficient manner? I can see there would be some terrible ways to do it, like iteratively increasing the pixel tolerance until picture.compare passes or summing together the RGB values of every pixel in a picture.difference result - which aren't really options.
