提出一种新的抗旋转攻击的图像水印算法.算法在小波域的高频细节图添加用于测量的基准线,在提取时通过哈夫变换检测出基准线估算旋转角度.根据估算角度,通过坐标变换在旋转图中搜索出原图的坐标,把像素值取出.由于取出了原图的非插值点,因此可以无失真恢复原图,进而无失真提取水印,此检测和校正算法不影响水印嵌入方法.结合混沌加密,选择在小波2、3层分解的高频系数嵌入水印信息,给出了整个算法的实现流程.
A new image watermark algorithm against rotation is proposed. This algorithm adds a datum line in the high frequency detail image in the wavelet domain. When the watermark extracted hough transform is used to detect the datum line and estimate the rotation angle. According to the angle, the original coordinate in the rotation image is searched for through coordinate transform to get the pixel. Getting the non-insert pixels from the original image enables a nearly lossless restoration of image and extraction of watermark. This method of estimation and restoration has no influence on the way of watermark embedding and extracting. It chooses to embed watermark message into the two and three level of the wavelet coefficient using chaos encryption and gives the realization of the whole algorithm.