An alternative method to pixel-by-pixel comparison for comparing a test image produced by the AUT against a baseline. Unlike pixel-by-pixel comparison, keypoint detection allows for positive matches on bitmaps that are not 100% identical to the baseline.
Keypoint detection is based on the creation a keypoint profile of all or a portion of the baseline image. The keypoint profile, which is generated by an image processing algorithm, is a set of points in the baseline, each of which has a keypoint descriptor which stores vectorized information about the image in the vicinity of that point.
During testing, keypoints are similarly generated for any AUT-produced images that are to be compared with the baseline. Positive matches are based on regions of the test image having keypoint profiles that are deemed similar to that of the baseline, based on a pre-established similarity threshold.