一种遮挡情况下运动车辆的跟踪算法

A tracking arithmetic for moving vehicle under occlusion

遮挡是基于图像对运动目标识别与跟踪时经常遇到的问题,也是动态图像处理较难解决的问题之一.针对这一问题,提出一种基于Kalman滤波和边缘匹配的跟踪算法.该方法通过当前帧目标边缘与实时更新模板的最优匹配来确定目标的位移量,首先采用Kalman滤波预测目标匹配搜索区域,然后在搜索区域利用边缘匹配精确定位目标,从而大大减少了匹配的计算量.实验结果表明,该算法对短时间内被遮挡的运动目标的跟踪和预测效果良好.

Occlusion often appears in recognizing and tracing the moving target based on image, which is a difficul problem to be solved in dynamic image process. Aiming at this problem, a tracking algorithm based on Kalman filtration and edge matching is present. The displacement vector is conformed by optimal matching of the target edge template and the current edge, and the template is updated in real time. The Kalman filter is used to forecast the search areas. The target is exactly localizated by matching the edge in the areas. It greatly reduces the computation of the matching. The experimental results show the algorithm is good for tracing and forecasting the moving target hided by other objects in a short time.