基于特征的弹性图匹配人脸识别算法改进

摘要

This paper presents a new method for optimization of Elastic Graph Matching(EGM) algorithm in face recognition.After face image graph is composed of feature vector extracted from 2D Gabor wavelet, labeled graph vector is obtained.The proposed algorithm adopts the Principal Component Analysis(PCA) and FLD algorithm reduces dimensionality of labeled graph vector.In comparison with the conventional methods, the proposed approach can obtain satisfactory results in the perspectives of recognition rates and speeds.%针对传统人脸识别弹性图匹配算法空间复杂度高、实时性较差的问题,提出一种弹性图匹配改进算法,将人脸图片特征点经Gabor小波预处理后,结合主成分分析(PCA)和Fisher线性判别方法(FLD)对生成的特征矢量进行处理,降低维数,减少计算量,同时在不降低识别率的前提下,提高识别速度.与传统的PCA算法、FLD算法、EGM算法进行仿真比较,证明该改进算法识别率高、实时性好.