基于SURF的NSCT域抗几何攻击水印算法  

NSCT-domain Watermarking Algorithm Robust to Geometric Attacks Based on SURF

作　　者：陈青[1] 姚绍华 陈祥[1] 

机构地区：[1]上海理工大学,上海200093

出　　处：《包装工程》2018年第1期162-167,共6页Packaging Engineering

基　　金：国家863计划基金(2012AA050206);上海自然科学基金(12ZR1420800);上海理工大学国家级项目培育基金(16HJPY-MS06)

摘　　要：目的针对常规图像水印算法抗几何攻击能力差的问题,提出一种基于SURF特征点和Schur分解的图像水印算法。方法将置乱后的水印嵌入到图像NSCT变换后进行Schur分解得到的上三角矩阵中,然后运用加速鲁棒性特征(Speed-Up Robust Features)算法提取图像低频子带中稳定的特征点和特征点描述符,利用特征点匹配对估计、校正含水印图像的几何攻击参数;利用RANSAC算法对特征点匹配对进行筛选,删除误匹配对,提高匹配精度,完成失真图像的几何校正。结果实验数据表明文中算法对几何攻击和常规信号攻击均有较好的鲁棒性,NC值平均达到0.9以上。结论文中的图像水印算法对几何攻击和常规信号攻击均具有较好的鲁棒性和不可见性,且提高了匹配精度。The work aims to propose an image watermarking algorithm based on SURF feature point and Schur decomposition with respect to the poor ability of the normal image watermarking algorithm against geometric attack. Firstly, the scrambled watermark was embedded into the upper triangular matrix obtained by the Schur decomposition after NSCT transform. Then, the stable feature point and feature point descriptor in the low-frequency sub-band of the image were extracted with the algorithm of Speed-Up Robust Features. The feature point matching pair was used to evaluate and correct the geometric attack parameter containing watermarked images. Secondly, the feature point matching pairs were screened with RANSAC algorithm to remove false matching pairs and improve the matching accuracy. The geometric correction of distorted images was completed. Experimental data indicated that the proposed algorithm had better robustness against the geometric attack and normal signal processing attack, and the average NC values were all above 0.9. The proposed image watermarking algorithm not only has better robustness against the geometric attacks and normal signal processing attacks with better invisibility, but also improves the matching accuracy.

关 键 词：SURF SCHUR分解 非下采样 CONTOURLET变换 几何攻击 

分 类 号：TS206[轻工技术与工程—食品科学]