一种基于分层模板的人脸检测方法

针对校园监控视频环境下的人脸检测中人流量大、侧面视频多、低清晰度等问题,提出了一种基于分层结构的脸部形态模板方法,该方法通过综合肤色标定方法、阴影检测算法和脸部分层次的形态特征检测算法,得出了人脸结构信息的4个控制参数.通过在Caltech人脸数据库、ORL人脸数据库和本实验室仿真人脸库的实验结果表明,本文所提出的基于分层结构的脸部形态模板方法能准确高效地检测出人脸,适合应用到校园视频系统中.

On a Face Detection Method Based on Layered Face Template

The problem of face detection in the school video system has been considered in this paper, such as high flow of people and profile video.A new face detection algorithm has been proposed on the basis of hierarchical morphological template method.The proposed algorithm combines the skin detection and shadow detection to obtain the feature of layered face.Furthermore, four parameters of face feature have been revealed.Comprehensive experiments on Caltech face dataset, ORL face dataset, and our lab face dataset show the robustness of the proposed method, which are suitable for school video system.