摘要
非线性人脸识别技术已经在技术上得到了较大的进步,数据集大小对识别方法的影响成为研究热点。针对大型数据集对线性和非线性人脸识别方法有何种不同的影响等问题,先后进行了类内变化对识别方法的影响和人数对识别方法的影响的一系列实验和讨论。研究结果表明随着人数的增加,线性方法的错误识别率呈线性增长,而非线性技术的误判率基本上保持不变化。在人数增多的情况下,识别率却产生降低的情况。并指出在研究大型数据集时,非线性方法具有一定的优势。
With the non-linear research on face recognition method, a data set size effect on recognition method has become the research hotspot. For large data sets on linear and nonlinear face recognition methods have different effects, within class variation on recognition method and the effects of the number of recognition was influenced by a series of experiments and discuss has been carried out. Research results show that with the increase in the number of linear method, the error recognition rate of linear growth. And nonlinear method for error recognition rate has re- mained stable, when the population increases with a downward trend. Nonlinear method has certain advantages in large data sets is point out.
出处
《科学技术与工程》
北大核心
2013年第1期201-205,共5页
Science Technology and Engineering
基金
2011年人工智能四川省重点实验室开放基金项目(2011RYY007)资助
关键词
人脸识别
线性
非线性
人数
face recognition linear Nonlinear number of population
