Optimal Image Watermark Using Genetic Algorithm and Synergetic Neural Network
Effective image watermark should meet some basic features, such as authentication, imperceptibility, robustness and security. Genetic algorithm is a kind of evolutionary optimization technique that can improve watermark imperceptibility and robustness. Through wellconnected fitness function with peak signal-to-noise ratio and normalized cross-correlation coefficient, the watermark sequence encrypted by two-dimensional chaotic stream encryption from a meaningful image is embedded into the DCT coefficients of host image through getting an optimal intensity by genetic algorithm. Synergetic neural network, offering a new and different approach to the construction of highly parallel structures for pattern recognition, is used in watermark identification to identify the extracted watermark and has the ability to recognize the original watermark quickly and accurately after attacks. The experimental results prove the validity of the optimal image watermark proposed in this paper.
digital watermark genetic algorithm synergetic neural network discrete cosine transform
Chen Yongqiang Peng Lihua
School of Computer Science Wuhan University of Science and Engineering Wuhan, China Muniment Room of Deanery Wuhan University of Science and Engineering Wuhan, China
国际会议
长沙
英文
2135-2138
2009-10-10（万方平台首次上网日期，不代表论文的发表时间）