一种层级处理的人脸光照归一化方法  

On a Hierarchical Processing Method for Facial Illumination Normalization

作　　者：黎利辉[1] 

机构地区：[1]广西民族师范学院数学与计算机科学系,广西崇左532200

出　　处：《西南师范大学学报（自然科学版）》2014年第5期105-111,共7页Journal of Southwest China Normal University(Natural Science Edition)

摘　　要：采用层级处理模式,提出了一种新的人脸光照归一化方法,先对图像光照进行自动判定,决定采用何种方法进行预处理,减小不同光照下人脸灰度的差异.再根据Gamma校正算法的特点,设计了一种更适应人脸图像校正的参数计算方法.本方法在样本归一化后,使得图像的本真信息不丢失,而且提高了图像对比度.大量的实验表明,该方法可以有效提高人脸识别功能的适用范围,能够在更远距离或复杂光照下进行识别.A new face illumination normalization method has been proposed based on hierarchical processing model to judges automatically the illumination of image ,and to decides w hich method is adopted for pre‐treatment ,so as to reduce the face gray differences under different illumination .According to the charac‐teristics of Gamma correction algorithm ,a more suitable method has been designed to calculate the correc‐tion parameters .After normalization by this method ,the facial features will not been lost ,and will im‐prove the contrast of image .A large number of experimental results show that this method can effectively improve the adaptive scope of facial recognition ,and can work under the conditions of longer distances and more complex illumination .

关 键 词：光照归一化 GAMMA校正 层级处理 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]