Block Normalization Based Blind Detectors For Spread Spectrum Watermarking Systems

In this paper, novel blind watermark detectors are presented lor spread spectrum watermarking systems. These detectors are based on the proposed block normalization method. The block normalization method reduces host signal interference by using the local statistics of each 8times8 discrete cosine transform block of the watermarked image. The simulation results show that the block normalization method considerably improves the bit error rate of the existing correlation, covariance and maximum likelihood estimation detectors.