Colour-range histogram technique for automatic image source detection

Abstract

Computer generated images are visually becoming increasingly genuine, due to advances in technology as well as good graphic applications. Consequently, making distinction between computer generated images and natural images is no longer a simple task. Manual identification of computer generated images have failed to resolve the problems associated with legal issues on exact qualification of images. In this work, a colour range histogram was developed to categorise colours in computer generated images and natural images from a point of reference. Four groups were selected, using the algorithm, consisting of exact Red-Green-Blue (RGB) code (group 1), colour code within a range of 10 (group 2), colour code within a range of 20 (group 3) and colour code within a range of 30 (group 4) from the point of reference. An optimised equation for the four Colour Code Groups (CCG) was developed. The computer generated images categorised an average of 69.8%, 92.9%, 96.9% and 98.6%, of any colour code for groups 1, 2, 3 and 4, respectively. The categorised colours for natural images were 31.1%, 82.6%, 90.8% and 95.0% for groups 1, 2, 3 and 4, respectively. The results showed that natural images contain a wide range of RGB colours which makes them different. Consequently, the disparity in the percentage of colours categorised can be used to differentiate computer generated images from natural images.