Blind detectors in spread spectrum watermarking for interference reduction

In this paper, we propose novel blind detectors for spread spectrum watermarking systems. There exists host signal interference in blind spread spectrum watermarking systems where host signal is not available to the end-users. Our aim is to reduce the interference since it limits the capacity of watermarking systems and increases the bit error rate of the recovered watermark even if there is no distortion or attack. The novel blind detectors, which are based on the proposed normalization method, is presented in this work. The proposed normalization method reduces host signal interference by using local statistics of 8x8 discrete cosine transform blocks of the watermarked signal. Simulation results demonstrates that the proposed method considerably reduce the bit error rate of the recovered watermarks in case of existing correlation, covariance and maximum likelihood estimation based detectors are employed at the receiver. Consequently, the proposed detectors reduce the host signal interference and increase the capacity of the system.