摘要
设计实现了基于粒子滤波器的实时人脸跟踪系统。系统以AdaBoost算法为人脸检测基础,以粒子滤波器算法为人脸跟踪基础,基于粒子离散度的控制机制以实现跟踪步长的自适应调整,同时,提出了基于HSV色彩空间的特征提取和匹配算法,以及基于2阶AR模型的人脸运动模型,有效提高了算法的跟踪精度和跟踪速度,并将跟踪结果作为人脸检测模块的反馈信号,增强了检测系统的目标捕获和目标校正能力。
A real-time face tracking system is designed in this paper. AdaBoost based algorithm is used to detect the face,and particle filter based algorithm is used to track the detected face. In order to increase the precision and speed of tracking system, three modifications are presented in this paper： controlling the dispersion of particles to adaptively adjust the tracking step,feature extraction and matching in HSV color space,and using 2nd order AR model to estimate the face moving model. Furthermore,the tracking result is used as a feedback to promote the system ability of object capture and verification. Experimental results show that the proposed system can detect and track the face accurately and efficiently.
出处
《电视技术》
北大核心
2007年第12期81-84,共4页
Video Engineering
基金
国家自然科学基金项目(60332030)
