摘要
局部二值模式LBP是一种有效的纹理描述算子,能够度量和提取图像的局部纹理信息,对光照具有不变性。针对LBP在人脸识别特征提取时存在维数过高、识别效果不佳的问题,在FERET人脸库上对LBP和Fisherface的结合算法进行实验。实验结果表明:相对于原始LBP表达方法,结合LBP和Fisherface的人脸表达能有效降低计算复杂度,同时也提高了识别精度。
Local Binary Pattern s（LBP） is an effective texture descriptive operator,which can measure and extract the local texture information.For the problem of the face recognition too high and the effect of recognition also poor when the feature vector extracted by LBP,We for carry out an experiment of using a method based on LBP and Fisherface on FERET database.The experimental results on FERET database demonstrate that compared with original LBP based implementation,our method can considerably reduce computational complexity while improving the recognition accuracy well.
出处
《软件导刊》
2012年第5期30-33,共4页
Software Guide
关键词
人脸识别
局部二元模式
主成分分析方法
线性判别分析方法
face recognition
feature extraction
local binary pattern
principal component analysis
Linear Discriminant Analysis
