Abstract: Aiming at the problem that traditional computergenerated image identification methods based on neural networks had the disadvantages of high difficulty in identification and low accuracy, we proposed a computergenerated image discriminant method based on wavelet transform. Firstly, the image wavelet features were extracted by decomposing twodimensional discrete wavelet transform, and the multidimensional wavelet feature vector was extracted by nlevel wavelet decomposition according to image wavelet feature. Secondly, the noise features of computergenerated image were extracted by a threedimensional transform domain wave denoising algorithm (BM3D). Finally, the support vector machine (SVM) classifier was used to identify the computergenerated image, the multidimensional wavelet feature and the noise feature were classified by the SVM classifier to solve the problem that the two features were merged to form a linear inseparable highdimensional feature, and the accurate identification of computergenerated image was realized. The experimental results show that the proposed method has higher accuracy and stability in the identification of computergenerated images.

Key words: wavelet transform, computergenerated image, identification method, support vector machine, highdimensional feature, SVM classifier