为保证图像数字水印的鲁棒性和安全性,结合加密技术,提出了一种使用协同神经网络的图像数字水印算法。该算法是先对有意义的灰度图像进行水印序列化处理后,再将其作为水印信号嵌入到载体图像的分块DCT系数直流分量里。水印的检测提取算法是采用协同神经网络,将疑似水印信号作为网络的输入,而网络的输出就是识别的结果。经仿真实验验证,该算法可承受一定的图像处理操作,不仅能同时完成水印的检测与提取,而且能有效判断水印的归属,具有较好的性能。

An digital image watermark algorithm is proposed based on the synergetic neural network,combined with encryption technique,to guarantee its robustness and security.The meaningful gray image is serially processed into watermark signal and embedded into the DC elements of the block DCT coefficient matrix of the host image.Watermark detection/extraction algorithm is realized by using synergetic neural network.The(network's) input is possible watermark signal and its output is the recognition result.The emulational experiments indicates that,the proposed algorithm can fulfill the watermark detection/extraction at one time after image processing,and identify the watermark owner which shows good performance.