CAMSHIFT结合HOMOGRAPHY变换的算法  

New Method to Extract Trajectory Based on CAMSHIFT Algorithm and HOMOGRAPHY Transformation

作　　者：李玉红[1] 崔国栋[2] 柴林燕[3] 

机构地区：[1]承德医学院计算机教研室,河北承德067000 [2]河北工业大学计算机科学与软件学院,天津300130 [3]中国医学科学院北京协和医学院生物医学工程研究所,天津300192

出　　处：《煤炭技术》2010年第8期212-213,216,共3页Coal Technology

摘　　要：提出了一种基于CAMSHIFT算法结合HOMOGRAPHY变换的球员轨迹提取算法。在求取球员位置的过程中,CAMSHIFT算法通过球员的颜色直方图模型实现对球员的跟踪。在跟踪过程中引入KALMAN滤波进一步提高算法的跟踪效率。最后利用HOMOGRAPHY变换将场景视频帧转换为俯视投影图,真实反映球员在球场的位置信息。实验结果表明该算法能够较好的提取出球员的运动轨迹。A new method to extract trajectory based on CAMSHIFT algorithm and HOMOGRAPHY transformation in soccer video is proposed in this paper.In the process of procuring the player＇s position,the CAMSHIFT algorithm is used to track the player based on the color histogram model.Kalman Filter is imposed to improve the tracking speed.Finally,Homography transformation is introduced to get the player＇s real position in the playfield.The experimental results show that the proposed method can extract accurate player＇s motion trajectory robustly.

关 键 词：跟踪 轨迹描述 CAMSHIFT HOMOGRAPHY KALMAN 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]