基于改进局部二值模式和梯度特征的计算机生成图像鉴别算法

针对目前计算机生成图像鉴别算法存在的计算复杂度高及检测率低等问题, 提出一种改进局部二值模式和梯度特征的计算机生成图像鉴别算法. 该算法主要基于图像的局部纹理特征, 先提取计算机生成图像和自然图像的特征向量, 再将该特征利用SVM分类器进行分类. 实验结果表明, 该算法可有效地鉴别计算机生成图像和真实图像.

Computer Generated Image Recognition Algorithm Based onImproved Local Binary Pattern and Gradient Features

Aiming at the problems of high computational complexityand low detection rate of current computer generated image recognition algorithm, we proposed a computer generated image recognition algorithm based on improved local binary pattern (LBP) and gradient feature. The improved algorithm wasmainly based on the local texture feature of the image. Firstly, the feature vectors of the computer generated images and the natural images were extracted, and then these features were classified by SVM classifier. The experimental results show that the proposed algorithm can effectively discriminate computer generated images and real images.