摘要
实现人脸图像识别真实性的过程中,为提升可变光照下人脸识别图像的识别率,提出基于小波变换的可变光照下人脸图像识别方法,小波分解人脸图像,过滤掉人脸图像中的冗余信息,获取人脸低频图像,采用Gamma校正改善图像亮度、通过高斯差分过滤出人脸图像低频中的光照和高频中的噪声,通过对比度变换增强图像质量,实现人脸图像的光照处理。采用小波变换和LBP编码方式提取光照处理后人脸图像的LBP直方图特征,计算特征间距离,近邻分类人脸图像,实现可变光照下人脸图像识别。经过实验检测发现,所提方法有效降低可变光对人脸图像的影响,完整保存人脸面部表情,并改善图像亮度对比度和滤出人脸图像中光照成分,识别可变光照下人脸图像的最高识别率达到98. 56%。
In the process of realizing the authenticity of face image recognition,in order to improve the recognition rate of face recognition image under variable illumination,a face image recognition method under variable illumination based on wavelet transform is proposed. The face image is decomposed by wavelet transform,the redundant information in face image is filtered out,the low frequency image of face is obtained,the brightness of face image is improved by Gamma correction,and the face image is filtered out by Gauss difference. The illumination in low frequency and noise in high frequency can enhance the image quality by contrast transformation to realize illumination processing of face image. Wavelet transform and LBP coding are used to extract LBP histogram features of face image after illumination processing. The distance between features is calculated,and the face images are classified by neighbors to realize face image recognition under variable illumination. The experimental results show that the proposed method can effectively reduce the influence of variable light on face image,preserve facial expression,improve image brightness contrast and filter out illumination components in face image,and the highest recognition rate of face image under variable light is 98.56%.
作者
高玉潼
雷为民
原玥
GAO Yutong;LEI Weimin;YUAN Yue(School of Computer Science&Engineering,Northeastern University,Shenyang 110169,China;School of Information Engineering,Shenyang University,Shenyang 110044,China)
出处
《激光杂志》
北大核心
2020年第1期118-122,共5页
Laser Journal
基金
国家自然科学基金(No.61401081)
关键词
小波变换
可变光照
低频图像
人脸图像
识别方法
直方图特征
wavelet transform
variable illumination
low frequency image
face image
recognition method
histogram feature
