复杂交通场景中的车辆检测与跟踪新方法

Method for detecting and tracking vehicles in complex traffic scene

提出了复杂交通场景中的车辆检测新方法,采用背景差法进行运动对象分割,并结合运动边缘检测以提高检测的准确性,对提取出的感兴趣区域按一定规则进行区域融合以检测车辆。提出基于区域特征匹配的车辆跟踪新方法,通过计算车辆链表中区域的特征向量与当前帧运动区域特征向量的距离确定匹配车辆,实现车辆跟踪。同时利用“确定车辆”、“临时车辆”规则解决车辆遮挡和阴影问题。实验结果表明该方法简单有效,能有效解决车辆遮挡和阴影问题,实时提取交通信息。

A new method for detecting and tracking vehicles in complex traffic scene is proposed. The method based on motion object segmentation by means of background difference and background subtraction from motion edge detection, identifies the vehicles through regional fusion of the interested regions which are respect to the vehicles or not in terms of some proposed rules. A new vehicle detection method based on regional characteristic matching is proposed, by which the distance between characteristic vectors of regions in vehicle list and the characteristic vectors of current frame motion regions can be calculated and used to match current motion regions and track the vehicles. Experimental results show that the method, with the aid of proposed concepts of 'confirmed vehicle' and 'temporary vehicle' for recovering momentarily missed vehicles or distinguishing obscured vehicles, can extract traffic information at real time.