摘要
Adaboost算法具有很好的实时性，但是也存在检测过程中鲁棒性不强，遇到遮挡问题检测失效等问题：针对这些问题，提出了基于改进Adaboost的人脸检测算法，该算法结合了CamshiR人脸跟踪算法并改进了原算法中的颜色直方图模型。以实际人脸检测与跟踪实验为例，证明了该算法在人脸自动检测跟踪过程中具有速度快、准确度高，能有效克服检测过程中遮挡以及类肤色干扰问题等。
Adaboost detection algorithm has the advantage of better real-time, but the algorithm also has bad robustness in tracking procedure, tracking failure when occlusion occurrence,etc. Aiming at these problems, presents the method of face detection based on Camshift algorithm, The algorithm is improved by integrating with Adaboost face detection and color histogram model. By actual face detection and face tracking as example are taken, this algorithm has advantages of quick speed and good accuracy, meanwhile overcome occlusion problem and disturbing by skin colored object in the process of detection.
作者
吴璇
李见为
肖坤平
王赞
WU Xuan, LI Jian-wei, XIAO Kun-ping, WANG Zan （Key Lab. of Opto-Electronic Technique of State Education Ministry, Chongqing University, Chongqing 400030, China）
出处
《电脑知识与技术》
2010年第4期2451-2452,2456,共3页
Computer Knowledge and Technology
基金
重庆市自然基金项目：基于语义流行学习的大规模人脸图像检索研究（CSTC2008BB2160）
