彩色图像多签名水印算法的研究

利用小波变换多分辨率特性和人眼视觉特征,提出了一种新的彩色图像多水印算法。算法先将不同强度的签名水印扩频到彩色图像不同色块的小波域,然后进行逆小波变换生成水印图像。多个有意义的签名水印能提供更可靠的版权识别,而且多水印相互配合使水印图像同时具有很好的透明性和鲁棒性。大量实验结果表明:该算法对常见的图像处理方法和压缩具有很强的鲁棒性,并且当彩色水印图像转化为灰度图像后仍能从中提取出水印。

Research on Multi-Signature Watermarking Algorithm for Color Image

Taking advantages of the multiresolution feature of wavelet transform and the property of human visual system, a multi-Signature watermarking algorithm for color image is presented here. With this algorithm, the multi-signature watermarks are mixed with the chaotic sequences and spread over the different wavelet domains of RGB color spaces of the image. The watermarked image is obtained by using the inverse wavelet transform. Multi-watermarking system can be a valid way of ownership identification, and can achieve good transparency and robustness. Experimental results and analysis are given to demonstrate that the proposed algorithm is extremely robust to common image processing such as noise addition, cropping, median filtering, and JPEG compression. Moreover, if the watermarked color image is converted into the gray image, the signature watermarks can still be detected.