Copyright authentication for images with a full counter-propagation neural network

Watermarking is a technique used for digital copyright protection. It has been developed to protect digital media from being illegally reproduced and modified. In this paper, a full counter-propagation neural network (FCNN) is applied to copyright authentication, where the ownership information (watermark) is embedded and detected by a specific FCNN. As opposed to traditional methods, the watermark is stored in the synapses of the FCNN. Because an FCNN has storage and fault tolerance, most attacks do not degrade the quality of the detected watermark image. Moreover, the watermark embedding procedure and detection procedure are integrated into the proposed FCNN. It accomplishes watermark embedding and detection of one or many watermarks using a multi-cover image. The experimental results show that the proposed method is more robust than Deng’s, Hsieh’s, Hsu’s, and Zhang’s methods under different attacks, and that it is imperceptible and authentic.