摘要
在视频处理领域的运动目标跟踪问题中,卡尔曼滤波器(KF)与扩展卡尔曼滤波器(EKF)已经得到了广泛的应用,但在复杂背景或是目标高机动运动的情况下跟踪效果并不理想。提出一种基于交互多模型算法(IMM),并采用去偏转换测量卡尔曼滤波器(CMKF-D)对运动目标进行跟踪的算法。该算法有效地解决了单一模型无法与运动特性相匹配的问题,并克服了KF、EKF对非线性模型线性化所引入的误差。以足球视频为例进行仿真实验,结果表明该算法有效地提高了视频序列中运动目标跟踪的准确率。
For tracking and measuring maneuvering target in sports video frequency , Kalman Filter（KF） and Extended Kalman Filter （EKF） has been widely used, but with low accuracy. A model that is combined with Interaction Multiple Model （IMM） algorithm and Debiased consistent Converted Measurements Kalman Filter （CMKF-D） algorithm is proposed for tracking and measuring the target in sports video frequency. It avoids the error that may be caused by transferring non- linear model to linear model through EKF and KF. The football video frequency simulation shows this algorithm can promote the tracking performance of maneuvering target in sports video frequency.
出处
《中国图象图形学报》
CSCD
北大核心
2009年第5期920-924,共5页
Journal of Image and Graphics
基金
国际合作基金项目(2007DFA20790)
