新型人脸检测算法在视频监控系统中的应用  

Application of New Face Detecting Algorithm in Video Surveillance System

作　　者：徐显日[1] 戴在平[1] 蔡灿辉[1] 

机构地区：[1]华侨大学信息科学与工程学院TI数字信号处理方案解决实验室,福建泉州362011

出　　处：《电视技术》2008年第4期89-90,96,共3页Video Engineering

基　　金：福建省科技计划项目(2005H034)

摘　　要：该算法基于Gentle Adaboost的多特征融合,首先通过运动和肤色特征提取人脸区域,再使用多层分类器定位人脸,最后在TMS320DM642 DSP上实现。实验表明,该算法的检测率达96%,速度也满足监控的要求。The algorithm is proposed by using multi-feature fusion of Gentle Adaboost. First, the face areas are extracted elementary based on the characteristics of motion and complexion. And then, the multilayer classifier is used to locate the face areas exactly. Finally, the algorithm is realized on TMS320DM642 DSP. The experimental results of the algorithm show that the detection rate is up to 96%, and the speed can also meet the requirements of surveillance.

关 键 词：人脸检测 边缘能量 视频监控 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]