基于SIFT变换和DWT-SVD的自适应图像水印算法

Adaptive Image Watermarking Algorithm Based on SIFT Transform and DWT-SVD

目的由于传统的自适应水印技术无法实现水印嵌入区域依据图像自身的特征而改变,且抵抗几何攻击的能力较弱,提出一种基于SIFT特征提取,结合DWT-SVD的水印嵌入方法。方法首先对图像进行SIFT特征提取,然后利用RANSAC算法选取出鲁棒性较好的特征点,通过特征点选取合适大小的水印区域,最后结合DWT-SVD算法将水印嵌入原始图像中。结果采用SIFT变换结合DWT-SVD算法嵌入的水印,含水印图像的视觉效果较好。含水印图像在没有受到攻击时,可以无损地提取出水印;在受到攻击时,提取出的水印品质较好,NC值均高于0.96,且算法的时间复杂度和空间复杂度均较小。结论实验证明,此方案在面对任意角度的旋转攻击,以及常规图像处理攻击有很好的鲁棒性,可以达到保护数字产品的目的。

Since the traditional adaptive watermarking technology cannot change the embedded area of the watermark according to the characteristics of the image itself,and its ability to resist geometric attacks is weak,a watermarking embedding method based on SIFT feature extraction combined with DWT-SVD is presented.Firstly,the SIFT feature is extracted from the image,then a feature point with good robustness is selected using RANSAC algorithm,a suitable size watermark area is selected according to the feature point,and then the watermark is embedded into the original image with DWT-SVD algorithm.The watermark embedded by SIFT transformation combined with DWT-SVD algorithm has a better visual effect.When the watermarked image is not attacked,the watermarks can be extracted losslessly.When it is attacked,the quality of the extracted watermarks is good,the NC value is higher than 0.96,and the time complexity and space complexity of the algorithm are small.Experiments show that this scheme has good robustness against rotation attacks at any angle and conventional image processing attacks,and can protect digital products.