SELF RECOVERABLE ADAPTIVE FRAGILE WATERMARKING SCHEME WITH TAMPER DETECTION AND RECOVERY

A self recoverable adaptive fragile watermarking scheme to detect and recover attacked parts with improved tamper detection ability is proposed in this paper. The Cover image is divided into 2x2 blocks and an adaptive watermark is generated from the quantized version of the 2x2 block. Instead of choosing the best possible values, the proposed watermark generation scheme divides the 2x2 quantized block into two 1x2 blocks to form the watermark. In order to aid in tamper detection, authentication bits are generated from the mapped block of the 2x2 block. The watermark bits (authentication bits and recovery bits) are embedded into the mapped block to form the watermarked image. The proposed watermark embedding scheme provides a quality PSNR of the watermarked image well above 35 dB. The watermarked image was tampered using different attacks like object deletion, object addition, change of content attack and addition of noises. The tampered blocks were detected using the proposed multi-level tamper detection scheme. The attacked parts of the watermarked image were reconstructed using the proposed tamper recovery scheme. The proposed tamper recovery scheme provides a quality PSNR well above 30 dB for various types of attacks. The performance of the proposed fragile watermarking scheme was also evaluated in terms of PSNR (peak signal to noise ratio), SSIM (Structural Similarity Index Measure), Probability of False Rejection (PFR) and Probability of False Acceptance (PFA) of the reconstructed image for various tampering ratios. The PFR and PFA Values of the proposed scheme are close to zero indicating that the tampered pixels are detected correctly.