基于CamShift和Kalman滤波混合的视频手势跟踪算法
基于CamShift和Kalman滤波混合的视频手势跟踪算法罗元，李玲，张百胜，杨红梅(重庆邮电大学智能系统及机器人研究所，重庆400065)摘要：提出了一种基于CamShift和Kalman滤波混合的跟踪算法，实现了对视频图像中动态手势的跟踪。在跟踪过程中，CamShift利用手势的颜色直方图模型，将图像序列通过一个肤色概率查找表转换为肤色概率分布图，结合运动信息和肤色概率分布，初始化一个搜索窗的大小和位置，并根据上一帧跟踪的结果自适应调整搜索窗口的位置和大小，从而定位出当前图像中手势的中心位置。在CamShift算法基础上利用Kalman滤波对搜索窗口进行运动预测。实验表明，该算法快速准确可靠，并且较好地处理了跟踪过程中大面积肤色干扰问题，对复杂场景的检测与跟踪也取得了较好的效果。同时，该算法还适用于其他具有特定颜色目标的跟踪。关键词：连续自适应数学期望移动；卡尔曼滤波；手势跟踪；颜色概率分布；搜索窗
中图分类号：TP391．4 文献标志码：A 文章编号：1001—3695(2009)03-1163—03
Video hand tracking algorithm based on hybrid CamShift and Kalman filterLUO Yuan，LI“llg，ZHANG Bai—sheng，YANG HoIIg-mei(Research Center ofIntelligent system＆Robotics，Chongqing University of Posts＆Telecommunications，Chongqing 400065，China)Abstract：Thi8 paper presented atracking algorithm based on hybrid CamShift and Kalman which implemented dynamic hand tracking in the video．In the process of object tracking，for each video frame，by using the object histogram model，CamShift converted the raw image to a flesh color probability distribution image via aflesh color probability table．Considering the hand motion and fiesh color probability distribution．initialized the size and location of a search window．Used computed information previously to adiust current search window’s size and location。then found the location of the hand，devised motion prediction of the search window using Kalman filter．Experimental results show that this algorithm is fast，accurate，and robust．It Call not only deal with the skin color interference problems．but also deal well with the track of complex background．And more，it tracks objects with special color．
Key words：CamShift(continuously adaptive lnean-shift)；Kalman filter；hand tracking；color probability distribution；search window
页:
[1]