一种稳健的人脸检测算法

A robust face detection method

人脸检测是人脸识别系统的重要组成部分,对于安全级别较高或特殊场合的门禁系统而言,高准确率的人脸识别技术尤为重要。为提高门禁系统的安全性,采用了多种特征相结合的人脸识别算法,融合了背景分离、肤色检测、人脸五官特征检测、运动物体轮廓分析、人体运动跟踪等多种技术进行人脸检测测试,有效地解决了单一特征的人脸检测方法对人脸进行漏检和误检的问题。实验结果表明,该算法在复杂背景和光照条件不足以及有遮蔽物的情况下,均能快速准确地检测出人脸,误检率低。

Face detection is a major part of face identification system. A robust and ettlctent atgorlmm tu, ,,~ particularly important for access control system which is applied in the situation of high security level or on some special occas- sions. In order to improve the security of access control system, a face detection algorithm combined with multi-characteristics is proposed in the paper and used to detect human face, which combines background separation, complexion detection, face features detection, profile analysis of moving objects, and moving objects tracking. And the proposed method solves the prob- lems of missing detection and false detection of human face caused by single-feature human face detection method. The experi- mental results indicate that the method detects human face fast and accurately with lower false detection rate in a complex background and in different lighting conditions.