基于光流和贝叶斯算法的目标跟踪研究

摘要

In this paper,a method is proposed to deal with the tracking of a moving target in the video stream and determine target velocity.It uses the optical flow and Bayesian boosting method to detect the target in every frame,which can improve detection performance of optical flow.In the process of tracking,the displacement of a moving target can be determined by calculating the pixel displacement of the centroid.The target velocity is computed by moving distance among frames.The results of experiment show that the method can realize the effective tracking of multiple moving objects.%　　在本文中提出了一种方法来处理视频流中移动目标的跟踪，并且能够确定目标的移动速度。利用光流与贝叶斯方法在每一帧图像中检测对象，这种方法能够提高光流的检测性能，利用目标的质心像素位移来确定目标的移动距离，目标的速度是利用系列帧中目标移动的距离之间的帧数来计算的。实验表明，该方法能够有效地进行目标跟踪。