Analysis of Perceptive Hashing Algorithm in Image Manipulation Detection

Content based image hashes are produced by a class of algorithms called perceptual image hashing. In comparison to cryptographic hashes, perceptual hashes have been designed to remain stable under minor modifications to an image including brightness, colour correction and compression. In order to generate a hash code, a perceptual hashing algorithm obtains robust features from an input image. This paper includes the study of robustness of the perceptive hashing algorithm on a large number of images under several content-preserving operations. We perform content-preserving operations on each image before computing the normalized hamming distance between input and transformed images. The experimental results show the robustness of perceptual hashing technique to a variety of content-preserving operations at various parameter values.