Finding unknown repeated patterns in images

We consider the problem of finding unknown patterns that appear multiple times in a digital image. We want to find the number of repetitions and their positions in the image without constraining the nature and shape of the pattern in any way. We propose a method that is able to pinpoint the possible locations of the repetitions by exploring the connection between image compression and image complexity. The method uses a finite-context model to build a complexity map of the image in which the repeated patterns correspond to areas of low complexity, which mark the locations of the repetitions.