基于NSCT_SVD_DE的自适应大容量水印算法

目的针对NSCT_SVD水印算法水印容量较小的问题,提出一种基于视觉区域特性和菱形编码的自适应大容量水印算法。方法对图像进行NSCT变换,对NSCT域低频系数矩阵进行8×8分块,选取每一块中最大的奇异值。根据人眼对图像亮度、纹理及平滑的敏感程度不同,选择在不同的分块嵌入不同大小的水印信息。结果实验表明,在6幅标准512×512像素的灰度图像中嵌入水印后的PSNR保持在39左右,具有较好的不可感知性,水印容量比DWT_SVD水印算法提升了约41.41倍,比NSCT_SVD水印算法提升了约20.95倍。对常见的中值滤波、高斯滤波、椒盐噪声、旋转和JEPG压缩等攻击具有较好的鲁棒性。结论该水印算法在满足不可见性和鲁棒性的前提下,增加了水印容量,具有一定的实用价值。

Adaptive Large-capacity Watermarking Algorithm Based on NSCT_SVD_DE

The work aims to put forward an adaptive large-capacity watermarking algorithm based on the visual regional characteristics and diamond coding, regarding the small watermark capacity of the NSCT_SVD watermarking algorithm. The NSCT transform was performed on the image, and the low frequency coefficient matrix of the NSCT domain was divided into 8 × 8 blocks, and the largest singular value in each block was selected. According to the different sensitivities of the human eye to the brightness, texture and smoothness of the image, the watermark information of different sizes was embedded in different blocks. The experiments showed that, the PSNR of the six watermarked grayscale images with 512×512 pixels was about 39, with a better imperceptibility, and the watermark capacity was about 41.41 times higher than that of the DWT_SVD watermarking algorithm, which was about 20.95 times higher than that of the NSCT_SVD watermarking algorithm. In the meantime, it had good robustness against the common median filtering, Gaussian filtering, salt and pepper noise, rotation, JEPG compression and other attacks. The proposed algorithm has increased the watermark capacity and has certain practical value on the premises of meeting the invisibility and robustness.