摘要
针对视频图像的数量海量性及视频人脸运动性,提出基于距离和密度聚类融合的视频人脸识别方法,该方法同时考虑距离与密度在聚类中的相关性并进行融合处理,设计了聚类人脸识别程序流程,实验中对其聚类算法及视频人脸识别方法分别进行了测试.结果表明,提出的方法识别速度快、准确率高、误检率低,能达到数字视频监控系统对精度的要求,可满足视频监控中实时取证的需要.
In allusion to video face movable and quantity magnanimity of video picture,a video face recognition method based on cluster integration of distance and density is presented,cluster relevance of distance and density is considered and integrated at the same time,the procedure of face cluster recognition is designed,the algorithms of cluster and methods of video face recognition have been tested respectively in the experiment,the result indicates,this method has some advantages of high speed of recognition,high rate of accuracy,low rate of mistake detection,being able to reach precision requirements of the digital videos monitoring system,and it can meet the need of the video controlling collection evidence in real time.
出处
《太原师范学院学报(自然科学版)》
2010年第3期56-59,69,共5页
Journal of Taiyuan Normal University:Natural Science Edition
基金
湖南省教育厅项目(08C027)
2007湖南省公安厅科研计划项目批准立项课题
关键词
视频人脸识别
聚类融合
距离
密度
实时取证
video face recognition
cluster integration
distance
density
collection evidence in real time