一种复杂光照条件下的人脸识别新算法
首发时间:2013-08-29
摘要:针对人脸识别中光照变化对识别率的影响,提出了一种光照条件变化下的正面人脸识别新算法。新算法融合了高斯滤波器、基于韦伯局部描述算子(Weber Local Descriptor,WLD)预处理方法、完备线性判别分析(Complete Linear Discriminant Analysis,CLDA)特征提取算法。首先进行高斯滤波与WLD算法预处理,提取出在变化条件下对光线不敏感的面部特征,然后选用CLDA算法进行特征提取,CLDA结合了零空间LDA(Null Space LDA,NS-LDA)和Fisherface方法,最后采用最近邻分类方法进行分类。利用受到不同程度光照影响的The Extended Yale Database B人脸库进行实验,并与直方图均衡化、局域标准化等预处理方法和经典的Fisherface、主分量分析、二维主分量分析等方法比较,新算法不仅具有较好的识别效果,且算法简便运行速度快,满足光照条件不理想的人脸识别实时性要求。
For information in English, please click here
A New Face Recognition Algorithm For Illumination Variation Environment
Abstract:Concerning the illumination changes on the influence of the recognition rate in face recognition,a new face recognition algorithm for varying illumination environment was proposed. The new algorithm integrated Gaussian filter, the preprocess method based on weber local descriptor, and feature extraction algorithm based on complete LDA for frontal face recognition. Before the feature extraction, firstly, Gaussian filtering and WLD preprocessing be used to extract the illumination insensitive representation of face images under varying illuminations, and then choose CLDA algorithm for feature extraction. Complete LDA integrate null space LDA and Fisherface algorithm. The nearest neighbor method was used to recognize the frontal face. Experimental results on the Extended Yale Database B face database show that novel algorithm performs better than the existing representative approaches such as HE、LN preprocessing method and classical Fisherface、PCA 、2DPCA algorithm. As its high recognition rate and fast superiority, it fully meet the real-time requirements for face recognition.
Keywords: artificial intelligence Weber's Law feature extraction
论文图表:
引用
No.****
同行评议
共计0人参与
勘误表
一种复杂光照条件下的人脸识别新算法
评论
全部评论0/1000