New fragile image watermarking schemes using chaotic sequences

Abstract

With the rapid development of digital signal processing tools, image contents can be easily manipulated or maliciously tampered with. Fragile watermarking has been largely used for content authentication purpose. This dissertation presents two new proposals for image fragile watermarking algorithms for tamper detection and image recovery. The watermarked bits are obtained from the parity bits of an error-correcting code whose message is formed from a binary chaotic sequence and from bits of the original image. In the first proposed algorithm, the watermarked bits are inserted in the frequency domain using the Discrete Wavelet Transform. The imperceptibility, detection, and recovery of this algorithm are tested for various attacks used in signal processing. In the second method, the watermarks bits are embedded using the least significant bit method. A comparison between the proposed algorithms shows that the former exhibits greater imperceptibility, while the latter exhibits better recover capability. The proposed algorithms are analyzed both for grayscale and color images. Comparison results reveal that the proposed technique performs better than some existing methods.