几何失真校正耦合位平面分解的图像水印算法

Image watermarking algorithm based on geometric distortion correction and bitplane decomposition

为提高当前图像水印技术的抗几何失真能力,兼顾其鲁棒性与不可感知性,设计稳定几何失真校正机制耦合重要位平面分解的鲁棒图像水印算法。基于离散小波变换DWT (discrete wave transform),对宿主数据的重要位平面实施处理,得到对应的子带信息;并把低通子带划分成一系列的子块,通过计算嵌入因子,构建水印嵌入机制,将加密水印隐藏到低通子带的子块中,形成水印图像;建立训练样本,提取这些图像的重要位平面,计算对应的低阶伪Zernike矩;利用训练样本对SVM (support vector machine)完成训练,校正失真的水印图像;设计水印检测方法,提取水印。测试数据表明,与当前图像水印方法相比,所提算法具有更强的抗几何变换能力,以及更好的不可感知性与鲁棒性。

To improve anti-geometric distortion ability of the current image watermarking technology, and considering both its robustness and imperceptibility, based on support vector machines, a robust image watermarking algorithm with stable geometric distortion correction mechanism coupled with important bitplane decomposition was designed. The important bit plane was decomposed based on the discrete wavelet transform to output a low pass band and other high pass subbands. The low-pass subband was divided into non-overlapping subblocks, and the intensity factor of watermark embedding was calculated to design watermark embedding method, and the watermark image was formed by taking the encypted watermark information into the non-overlapping subblock of the low pass band. A training sample was set up to extract the important plane of these images, and the corresponding low order pseudo Zernike moments was calculated. The support vector machine was trained based on the training samples to correct the geometric distortion of the watermark image. The watermark detection method was designed for accurately extracting the watermark. The test data show that the proposed algorithm has stronger ability of resisting geometric transformation, as well as better invisibility and robustness compared with the current image watermarking method.