摘要
光照是影响人脸识别率的关键因素,它已成为人脸识别技术发展的瓶颈。在光照变化对人脸图像影响分析的基础上,按照处理方式的不同对光照处理方法进行了分类和评价。从基于不变特征的方法、光照变化建模的方法、人脸图像归一化的方法和基于SFS的方法等四个方面介绍了相关的算法和理论,分析了各种方法的优缺点,并提出了关于人脸光照问题的进一步研究方向。
Illumination is a key factor in face recognition. It is still one bottleneck in the development of face recognition technique. Based on the analysis of illumination changes on face images, illumination treatment approaches are introduced and classified according to different processing modes. The associated algorithms and theories are presented from the four aspects, including invariant features-based method, illumination change modeling method, canonical model method and shape-form-shading model method. After analyzing advantages and disadvantages of these algorithms, directions of the future research are given.
出处
《电脑开发与应用》
2008年第4期36-39,共4页
Computer Development & Applications
基金
陕西理工学院重点科研基金项目(SLG0630)资助
关键词
人脸识别
光照
商图像
光照锥
球谐函数
face recognition, illumination, quotient image, illumination cone, spherical harmonics
