文中主要研究封闭区域内多摄像机间车辆跟踪时车辆的同一性识别问题,提出一种基于车辆多种特征的、多摄像机间的车辆目标一致性匹配方法.该方法采用背景减除法分割出感兴趣区域,对感兴趣区域采用灰度和彩色区间分割相结合的方法定位车牌,并识别车牌信息及车辆的HSI颜色信息;同时利用边缘检测算法和矩不变量技术对车型进行识别,最后采用特征加权法对车辆进行一致性匹配.实验结果表明这种匹配方法在外界条件良好时可以较好地检测出同一车辆目标.%It presents a method of target consistency match of vehicles based on multi-features and different cameras. With the help of background minus division,it picks up regions of interest, which could lead to the recognition of license plate by the combination of breakup of grayscale and colorful zones. And the information of HIS about the vehicle is obtained, too. The algorithm of edge detection and the technology of invariant movements are applied to distinguish the vehicles. Eventually, the consistency of the vehicles is judged by weighted features. Experimental results show that vehicle target can be detected when have a good external conditions.
展开▼