一种基于SURF特征点校正的二维码图像水印算法

A Two-Dimensional Image Watermarking Algorithm Based on SURF Feature Point Correction

针对图像水印存在的容量小,抗几何攻击能力差等问题,提出将二维码作为水印信息,嵌入DCT(离散余弦变换)域中。提取水印时首先利用SURF(Speed-Up Robust Features)提取特征点,在计算变换参数之前规定特征点区域,提高特征点的匹配正确率,最后利用最小二乘法计算仿射变换矩阵,通过逆变换使受攻击水印和图像恢复坐标同步,提高水印检测的可靠性。理论分析及结果表明,该算法不可见性良好,水印容量大,并能抵抗旋转、缩放和平移等常规几何攻击。

In order to solve the problems that the capability of image watermarking is small and the anti-geometric attack abili-ty is poor,QR Code is proposed as the information of watermarking and embedded in the DCT(Discrete Cosine Transform)domain.The characteristic points of watermarking are extracted by SURF(Speed-Up Robust Features)at first. Secondly,it should deter-mine the region of feature points before calculating the transformation parameters to prove the correct rate of matching feature points.Finally,it used the least squares method to calculate affine transformation matrix. According to Inverse transformation to recoverycoordinates of the watermarking being attacked and images to improve the reliability of testing watermarking. The theoretical analy-sis and the results show that the Invisibility of algorithm is good,the capacity of watermarking is large,and it can resist geometric at-tacks of rotating,scaling and translation.