Abstract: This paper presents an algorithm detecting differences between computer generated figure images（CG） and figure photographs（PG）.First obtain a statistical model by extracting wavelet coefficients features，color and texture features，and then SVM classifiers are used to study and identify this difference.Experiments have shown that the method in this paper has a certain reference value in identifying the authenticity of images，especially in the conditions that there are not digital watermarks in the image.The method is of special practical value and its features of the extraction is significantly effective，and it has been verified that the combinding of statistical model and classification is effective in the image recognition authenticity.

Key words: image Recognition authenticity, Photograph, figure　image, wavelet, gray-level co-occurrence matrix, Support Vector Machine

摘要： 提出了一种区分计算机绘制的人物图像（以下简称CG图像）和由摄像机获取的真实人物图像（以下简称PG图像）的方法，该方法通过提取小波系数特征、颜色特征以及纹理特征获得图像的统计模型。并采用SVM分类器来学习和辨别这种差别。实验表明，该方法在分辨图像真伪方面是具备一定参考价值的，尤其在一幅图像没有数字水印的情况下，具有特殊的实际应用价值。其对特征的提取是显著有效的，从而验证了这种统计模型与分类器相结合进行图像真伪分析的实验方法的有效性。

关键词: 图像真伪, 人物图像, 真实人物图像, 小波, 灰度共生矩阵, 支持向量机