摘要
为了提高光照条件下的人脸识别正确率,提出一种复杂光照条件下的人脸预处理算法。对人脸图像进行局部增强处理,用双边滤波对图像亮度进行估计,采用Gamma校正补偿图像亮度估计产生的损失,将反射分量与亮度估计结果融合获得效果更优的人脸图像,并用K近邻算法建立分类器对人脸进行识别。在Yale、PIE和AR人脸库仿真结果表明,该算法提高了光照条件下的人脸识别正确率,其性能优于当前典型人脸识别算法。
In order to improve the face recognition rate in illumination conditions, this paper proposes a face preprocessing algorithm in complex illumination conditions. The illumination face image is enhanced by nonlinear method, and then bilateral filtering is used to estimate the influence of illumination, and the loss of estimation is compensated by Gamma method.The estimation results are fused to improve the effect of face image, and face classifier is established based on K near algorithm. The results of Yale, PIE and AR face database show that the proposed method has improved face recognition rate in complex illumination conditions compared with other methods.
出处
《计算机工程与应用》
CSCD
2014年第19期187-191,共5页
Computer Engineering and Applications
基金
湖南省教育厅科研项目(No.12C0681
No.13C336)
关键词
人脸识别
支持向量机
亮度补偿
图像增强
face recognition
support vector machine
illumination compensation
image enhancement