Abstract
We present a tamper detection algorithm based on cat map and discrete wavelet decomposition. The algorithm is fragile to any tampering modification, but it is robust to harmless common image processing operations like JPEG compression, resizing, noising and filtering. The detection algorithm generates a witness image showing the tampered regions based on the blind analysis made on the tampered image. The embedding algorithm uses the approximation coefficient of a m x m block image as the watermark to be inserted in the details coefficients of another block. The blocks pairs are associated using a cat map permutation of m x m blocks. Moreover, we have been able to recover most of the original watermarked image based on a threshold depicted from the witness image. Simulations results demonstrate the efficiency of the tamper detection and recovery algorithm. We use many metrics to quantify the imperceptibility level like the PSNR, wPSNR, UIQ and SSIM. Also sensitivity and false alarm levels of the detection algorithm are measured and reported.