摘要
传统LBP算法通常采用统计直方图特征来表达人脸图像信息,故无法包含空间结构信息。针对此问题,文中提出一种基于局部二值模式(LBP)和马尔科夫稳态特征(MSF)的人脸识别方法。该算法避免了上述传统LBP算法的缺陷,使得LBP直方图特征之间的相对位置信息得以保留。在此基础上引入分块统计LBP-MSF的方法,从而使统计特征能体现人脸整体结构信息。在FERET标准人脸数据库上的实验结果表明该算法对于提高人脸识别率非常有效,可得到优于传统LBP算法的最高96%人脸识别率。
The original LBP algorithm usually utilizes the histogram-based features as the facial feature information,so it contains no spatial structure information. In order to address this problem,a new face recognition algorithm is proposed in this paper,which is based on the Local Binary Pattern（ LBP） and Markov Stationary Features（ MSF）. This algorithm can alleviate the limitation of the original LBP algorithm and make the spatial location information be preserved in the LBP histogram-based features. Additionally,we also apply the block division method to extend the LBP-MSF algorithm so as to make the extracted facial feature representation contain the whole spatial structure information. Experiments are conducted on the FERET face database,the maximum recognition accuracy of96% can be obtained which is better than the original LBP algorithm. The experimental results show that our proposed algorithm is very effective for face recognition.
作者
吴学谦
李菲菲
颜艳
陈虬
WU Xueqian;LI Feifei;YAN Yan;CHEN Qiu(Electrical and Computer Engineering,University of Shanghai for Science and Technology,Shanghai 200093,China)
出处
《电子科技》
2018年第8期14-16,20,共4页
Electronic Science and Technology
基金
上海市高校特聘教授(东方学者)岗位计划(ES2014XX
ES2012XX)
关键词
局部二值模式
空间结构信息
马尔科夫稳定特征
人脸识别
local binary pattern
spatial location information
Markov stationary teatures
face recognition
