Blind Multipurpose Image Watermarking with Perfect Security

Abstract

Nowadays, from one hand multimedia authentication techniques are widely used to achieve trustworthiness, on the other hand, due to the rapid growth of image processing software technologies, having a secure method to protect the copyright of these data seems fairly essential. Multipurpose watermarking emerged in order to simultaneously accomplish multimedia authentication and copyright protection. In this paper, we propose a multipurpose watermarking method which achieves perfect security, the ability to detect tampered areas of the watermarked image as well as a lower BER rate, at the cost of reducing capacity by half. This watermarking scheme is blind in the sense that on the receiver side, neither the original host image nor the embedded watermark is needed for ownership watermark extraction or tamper detection. Experimental results show that our method is able to reconstruct extracted tampered watermarks even after various attacks such as JPEG compression, average filtering, gamma correction, median filtering, speckle noise, JPEG compression, sharpening, Wiener filter, and median filtering. Comparisons are provided with other multipurpose watermarking methods which primarily aim at simultaneous goals of copyright protection and authentication. We also show the superiority of our proposed method to three watermarking methods attaining these objectives on a one-goal-at-a-time basis.