A finite field cosine transform-based image processing scheme for color image encryption

In this paper, we propose a color image processing method based on a finite field cosine transform (FFCT). Basically, the approach consists in applying a 32-point FFCT over GF(224) to blocks of a color image. Since RGB images are considered, each pixel of the corresponding image blocks can be represented as a binary 24-tuple and then treated as an element of GF(224). By means of computer experiments, we show that such a joint three-channel processing leads to the uniformization of the histograms of an image and decorrelates adjacent pixels. This makes our method useful to provide robustness against statistical attacks in image encryption schemes.