基于Kalman的运动物体跟踪和预测方法研究

目前,很多应用需要跟踪图像序列中的运动物体。但是,有时不知道运动物体的特性,因此,提出一个完整的跟踪预测模型;使用无需先验知识的Kalman滤波器跟踪和预测运动物体。利用提取的Harris角点,通过L-K金字塔方法得到前后两帧光流;通过光流聚类得到当前帧中运动物体的凸包,使运动物体从背景中分离出来。由Kalman滤波器跟踪和预测各运动物体凸包的重心,并划出运动轨迹。计算机仿真及现场测试结果表明所提出的方法具有较高的跟踪精度,且计算量小。

Using Kalman to Research Tracking and Prediction of moving object

Currently, many applications require tracking moving objects in image sequences. However, sometimes we do not know the characteristics of the movement, therefore, this paper puts forward a complete track forecasting models, using Kalman filter to track and predict the movement of objects without prior knowledge. Useing the extracted Harris corner to obtain optical flow between two frames by L-K pyramid method, getting the current frame of the convex hull of a moving object by optical flow clustering to separate the moving objects from the background. Tracking and predicting the movements of objects convex hull of gravity. Computer simulation and field test results show that the proposed method has higher tracking accuracy, and small amount of calculation.