以回归型支持向量机（SVR）理论基础，提出了一种可有效抵抗几何攻击的图像水印检测新算法.该算法首先选取图像的组合矩作为特征向量,并通过SVR对旋转、缩放、平移等几何变换参数进行训练学习，以获得SVR训练模型；然后利用SVR训练模型对待检测图像进行数据预测，并结合预测输出结果对其进行几何校正；最后从已校正数字图像内提取出水印信息.仿真实验结果表明，本文算法对常规信号处理（滤波、叠加噪声、JPEG压缩等）和几何攻击（旋转、缩放、平移、剪切等）均具有较好的鲁棒性。

In this paper, a robust image watermarking detection based on support vector regression (SVR) is proposed. Firstly, six combined low order image moments are taken as the feature vector and the geometric transformation parameters are regarded as the training objective, the appropriate kernel function is selected for the training, and a SVR training model can be obtained. Secondly, the combined moments for test image are selected as input vector, the actual output is predicted by using the well trained SVR, and the geometric correction is performed on the test image by using the obtained geometric transformation parameters. Finally, the digital watermark is extracted from the corrected test image. Experimental results show that the proposed watermarking detection algorithm is not only robust against common signals processing such as filtering, sharpening, noise adding, JPEG compression etc, but also robust against the geometric attacks such as rotation, translation, scaling, cropping, combination attacks, etc.