摘要： 为提高人脸识别对人脸姿态、位置、表情变化的鲁棒性，提出一种基于非下采样Contourlet变换(NSCT)与改进脉冲耦合神经网络(M-PCNN)的人脸特征提取方法。利用NSCT对输入图像进行多尺度分解和多方向稀疏分解，以捕获图像中的高维奇异信息，使用M-PCNN模型提取各子带的信息熵，将其作为人脸特征，利用支持向量机(SVM)实现分类与识别。仿真结果表明，该方法鲁棒性较强，在识别和分类中表现出较好的性能。

关键词: 人脸识别, 主成分分析, 保局投影, 特征提取, 信息熵, 支持向量机

Abstract: In order to improve the robustness of face recognition to the changes of facial pose, position and expression, a facial feature extraction method based on Nonsubsampled Contourtlet Transform(NSCT) and Modified Pulse-coupled Neural Network(M-PCNN) is proposed in this paper. By using NSCT, the input images are decomposed into a number of sub-images with various scales and directional features. The different subbands are decomposed into a sequence of binary images by using M-PCNN. The information entropies of each binary images are calculated and regarded as facial features. A Support Vector Machine(SVM) classifier is employed to implement recognition and classification. Simulation results show that this method has good robustness, and can achieve better result in verification and classification.

Key words: facial recognition, Principal Component Analysis(PCA), Locality Preserving Projection(LPP), feature extraction, information entropy, Support Vector Machine(SVM)

中图分类号: 

• TP391