Abstract: As the improvement of the social public safety system, smart video surveillance technology based on face plays a big role in safety monitoring, video analysis and man-computer interaction systems, etc. Although the traditional Camshift algorithm can track the moving object quickly, it not only needs to set the tracking object by manually, but also fails to track the object easily while it is occluded and interfered by the same color obstructions. To resolve above-mentioned problems, this paper combines Adaboost, Camshift and Kalman filtering algorithm which based on the OpenCV to realize face detection and tracking automatically and accurately.

Key words: OpenCV, Adaboost algorithm, Camshift algorithm, Kalman filtering, face detection and tracking

摘要： 随着社会公共安全体系的逐步完善，基于人脸的智能视频监控技术在安全监控、视频分析以及人机交互等场合发挥出越来越重要的作用。传统的Camshift算法虽然能快速地跟踪运动目标，但它不仅需要手动设定跟踪的对象，而且当跟踪对象遇到遮挡和相同颜色障碍物干扰时很容易丢失目标。针对上述问题，在OpenCV的基础上，采用Adaboost，Camshift和Kalman滤波相融合的方法，实现了快速、自动和准确的人脸检测与跟踪。

关键词: OpenCV, Adaboost算法, Camshift算法, Kalman滤波, 人脸检测与跟踪