摘要
基于Viola-Jones框架的面部识别算法,提出使用矩形特征法进行特征值运算,采用感知器学习算法训练最佳特征,确定最佳阈值的算法.实验结果表明:该算法运算速度较快,错误分类率较低,识别率较高,误识率较低.
A face recognition algorithm is illuminated Viola - Jones frame and it is ameliorated to improve the speed of eigenvalue calculation and to reduce the wrong recognition rate. The methods of PM is used to train the best character and find the best limen. It also uses the ameliorated expand - algorithm. Thus the wrong recognition rate is much lower. Through the real datas experiment, it shows that the result is better.
出处
《东北师大学报（自然科学版）》
CAS
CSCD
北大核心
2005年第3期24-27,共4页
Journal of Northeast Normal University(Natural Science Edition)
基金
国家自然科学基金资助项目(20093937)
关键词
面部识别
识别率
误识率
识别速度
face recognition
recognition rate
wrong - recognition rate
recognition speed
