中文摘要：

在签别伪图像问题的研究中,随着图像处理技术的提升,计算机合成的图形越来越逼真,如何准确区分自然图像和计算机图形,成为图像认证研究的重要内容。由于利用自然图像和计算机图形在高阶统计特性上的不同的特点,提出一种新的高阶统计特征与预测误差矩阵相结合的分类鉴别方法。利用三级正交镜像滤波器（QMF）提取图像的各级分量,并求出各级分量及其预测误差矩阵的高阶累积量作为特征数据,然后利用支持向量机（SVM）进行训练和鉴别。实验结果表明对于实验所用的图像库具有99.10%的高鉴别率,能够有效鉴别自然图像和计算机图形,同时方法复杂程度较低、具有良好的鲁棒性及稳定性。

英文摘要：

With the advance of image processing technology,computer graphics becomes more realistic.How to distinguish natural images and computer graphics accurately becomes an important part of image detection technology.A novel approach based on higher-order statistics and prediction-error matrix was proposed to detect natural images and computer graphics,according to statistical differences.First,the image decomposition employed here was based on a 3-level quadrature mirror filter（QMF） and components of three levels are extracted.Then,prediction-error matrix and higher-order cumulate were used as the feature data.Support Vector Machine was chosen as a classifier to train and test the given images.Experimental results demonstrate that this new detection scheme has a high identification rate of 99.10%,with less complexity,robustness and good stability.