摘要
该文提出一种在小波域内自适应于图像内容的数字水印方案.算法利用图像系数的零树结构,自适应地确定被嵌水印序列的长度,并根据小波域的量化噪声自适应地确定水印的嵌入强度,保证了水印序列的能量小于图像可容纳的噪声上限,使水印图像同时具有良好的视觉效果和鲁棒性.在检测端可根据水印图像的统计特性确定水印的检测阈值,不需要参考原图就能判断水印是否存在.对水印图像的攻击实验表明该算法对噪声干扰及常见的图像处理技术具有较好的鲁棒性.
This paper addresses the issue of how to determine the capacity and strength of the embedded watermark according to the image's content. With the approach proposed in this paper , the length of the watermark sequence is determined by the size of significant coefficients, which are selected by exploiting the tree-based spatial-frequency feature of wavelet transform. The watermark bits are inserted in the high activity texture regions with the maximum strength of Just Noticeable Distortion (JND) tolerance of the Human Visual System (HVS). So the length and strength of the embedded watermark are both image-adaptive, which greatly improves the perceptual quality and robustness of watermarked image. Through analyzing the statistical property of the watermarked image, the detection threshold is determined according to the false alarm and missed detection rate, the embedded watermark can be blindly detected by the correlation value between the marked image and reference watermark sequence. Simulation results show that the proposed method can achieve a better compromise between the robustness and transparency.
出处
《计算机学报》
EI
CSCD
北大核心
2002年第11期1195-1199,共5页
Chinese Journal of Computers