Cryptanalysis on digital image watermarking based on feature extraction and visual cryptography

Abstract

This paper presents the cryptanalysis on a method that uses the visual cryptography and feature extraction for watermarking in images. Many of the existing digital watermarking techniques use a method which does not involve embedding of the watermark into the vulnerable images, but rather it only uses the features of image as the security key for generating watermark. This is assisted with the visual cryptography which makes it possible to perform watermarking without embedding any data onto it. Although they perform very well with many different attacks like scaling, rotating, compression, different kind of noises, etc., it still fails to prove the copyrights of image when the attack as illustrated in the paper is applied to the watermarked image. This method allows the attacker to use his own valid watermark to be introduced to the image, which will result in failure of the owner to prove its copyright.