Perceptual Hash Computation of Multimedia Objects Using Improved KL Transform

This paper presents an improvement for Karhunen–Loeve (KL) transform which is a new approach for computation of perceptual hash in multimedia objects. Basically, all the cryptographic algorithms used for computation of hash function for multimedia objects suffer from avalanche effect. Perceptual hash function is a solution to the avalanche effect problem. To achieve the goal, the original image size is reduced into 8 × 8 pixels which in turn divided into sixteen 2 × 2 matrices (after grayscale conversion). The prime benefit of proposed algorithm is to improve the complexity of the 2D KL transform and also simplify the structure which can give a chance for recursive and parallel processing of images.