摘要
人脸识别问题在很多情况下都会面临小训练样本的问题,在训练样本数量远小于数据维数的情形下许多人脸识别方法都会遇到困难。本文分析了造成小样本问题的原因,从虚拟样本扩充、数据降维以及算法优化等不同方面总结了解决方法,并对不同方法进行了简要评价,对解决小样本问题的未来发展方向进行了展望。
Small Sample Size Problem is a general problem in face recognition. Most methods of face recognition will encounter difficulties when the number of samples is much smaller than the dimension of face data. This paper analyzed the the origin of the problem, and then defferent solutions were presented includeing virtual sample expansion,data reduction and algorithm optimization. A brief evaluation of different methods was given. Future directions of solving the problem were prospected .
出处
《软件》
2014年第3期167-169,共3页
Software
基金
北京市属高等学校人才强教计划资助项目(PHR201108261)
国家自然科学基金资助项目(61170116)
关键词
人脸识别
小样本问题
虚拟样本
数据降维
Face Recognition
Small Sample Size Problem
Virtual Sample
Data Reduction
