一种基于图象特征的小波水印

本文提出了一种用于图象版权鉴定的小波水印方法。该方法在一定的量化标准下利用小波变换提取图象的特征信息，通过混沌序列对其加密生成数字水印。然后，根据HVS的特点和小波系数间的内在联系确定嵌入强度和位置，将水印信息自适应地嵌入到低频部分。该方法保证了鲁棒性，并实现了脱离原图和参考水印的高概率检测。实验结果表明，这是一种行之有效的图象鉴定方法。

A Wavelet Watermark Based on Image Features

This paper proposes a watermarking method based on wavelet transform for copyright authentication of images. In this method, the original image' s feature is extracted by using wavelet transform and quantization . The feature information is changed into watermark by the chaotic sequence encryption. According to the characteristics of human visual systems(HVS) and the intrinsic relations of wavelet coefficients, the embedding intensity and places are determined. Then the watermark is adaptively embedded into the low frequency part. A secure detection algorithm without the original image and the reference watermark is developed in this paper. Thus, the robustness and high probability detection ability are ensured. Experiments show the method is very effective for image authentication.