基于Kalman滤波器与肤色模型的手势跟踪方法  

Hand Gesture Tracking Method Based on Kalman Filter and Skin Color Feature

作　　者：覃文军[1] 杨金柱[1] 王力 赵大哲[1] 

机构地区：[1]东北大学医学影像计算教育部重点实验室,辽宁沈阳110819 [2]公安部沈阳消防研究所,辽宁沈阳110034

出　　处：《东北大学学报（自然科学版）》2013年第4期474-477,共4页Journal of Northeastern University(Natural Science)

基　　金：国家自然科学基金资助项目(61001047;61172002);中央高校基本科研业务费专项资金资助项目(N110804005)

摘　　要：有效和鲁棒的手势跟踪是动态手势识别的前提,针对手势及其运动的特点,提出了结合Kalman滤波器和肤色模型的手势运动目标跟踪方法.首先通过背景差法和YCb’Cr’空间上的椭圆肤色模型检测出手部运动目标,通过目标区域的空间结构参数来设置Kalman滤波器的各项运动参数,然后计算空间结构特征的跟踪匹配函数对目标预测位置进行修正,获得运动手势目标区域及其运动轨迹.实验结果表明,所提方法能有效地跟踪手势,并能较好地适应手势在运动过程中的手形变化、轨迹转弯等情况,检测准确,鲁棒性高.The effective and robust hand gesture tracking is the premise of the dynamic hand gesture recognition. According to the characteristics of hand gesture movement, a hand gesture tracing method was presented on the basis of the Kalman filter and the skin color model. Firstly, the hand movement target was recognized with the background difference and the skin color model. At the same time, the movement parameters of Kalman filter were set by the parameters of spatial structure. Then, the tracking matching function based on the feature of spatial structure was calculated to correct the target predicted position, and the target region of dynamic hand gesture and its motion track were also got. The results showed that the proposed method was effective, accurate and robust to track hand gesture and was adapted to hand change, turning path and other situations.

关 键 词：手势识别 目标跟踪 KALMAN滤波 肤色模型 动态手势 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]