提出一种基于小波变换和QR分解的图像脆弱水印方法。嵌入水印时,首先对原始图像进行三级小波分解,然后提取第三级低频子带的边缘特征,同时对此边缘特征和第三级垂直高频子带进行QR分解,用边缘特征的一部分分解结果替换垂直高频子带分解结果的相应部位未嵌入水印,最后进行小波逆变换得到嵌入水印的图像。图像认证时,首先对观测图像进行三级小波分解,第三级垂直高频子带与低频子带边缘特征的QR分解结果进行比较,就可以准确认证图像是否经过攻击,并可以精确定位受攻击的部位。实验结果表明此方法的有效性和可行性。
A fragile watermarking algorithm based on wavelet transform and QR decomposition was_proposed. When embedding watermarking, the original image was decomposed by 3-level discrete wavelet first, and then the edge characteristic of the third approximation coefficient was extracted. The watermarking was embedded by replacing a part of Orthogonal-triangular decomposition results of the third vertical detail coefficient by those of edge characteristic. In the end, the approximation coefficient, changed detail coefficients were constructed to obtain image having watermarking. When authenticating image, the observed image was decomposed by 3-level discrete wavelet first. Image can be authenticated correctly and attacked parts can be localized through comparing the Orthogonal-triangular decomposition results of the third vertical coefficient with those of edge characteristic of the third approximation coefficient. Experimental results show the efficiency and feasibility of this method.