Self-embedding and Variable Authentication Approach for Fragile Image Watermarking Using SVD and DCT

Abstract

AbstractIn this paper, we propose a self-embedding fragile image watermarking technique based on Singular Value Decomposition (SVD) and Discrete Cosine Transform (DCT). To improve security and robustness a novel block separation technique is presented in which an input image is divided into non-overlapping blocks and subsequently SVD is applied on each block. Entropy value of the resulting vector of singular values is measured. Mean of entropy of all blocks is considered as a threshold for deciding two sets of blocks. For the first set of blocks, Entropy measures of singular value based authentication codes are generated and for second set of blocks, DCT DC value based authentication codes are obtained for embedding information. To enhance visual quality of watermarked and recovery image DCT based watermark embedding technique is proposed. Experiments have been conducted on gray scale images and it is observed that proposed scheme gives peak signal to noise ratio (PSNR) of 58 dB which ensures high performance in terms of imperceptibility. Experimental results showed that 99% of watermarked blocks are correctly detected during recovery phase. The image tampering is precisely identified and tampered images are recovered with very good quality. To the best of our knowledge the presented fragile image watermarking scheme is superior to all existing scheme.KeywordsAuthentication codeBlock separationDCTEntropySVDSelf-embeddingFragile