基于车辆轮廓定位匹配的车型识别方法

采用背景减法提取运动车辆目标的轮廓,用外接矩形对目标轮廓进行定位,选取轿车、客车和货车3种车型的侧面轮廓作为标准样本,分别计算待识别车辆的侧面轮廓与3种标准样本的匹配程度系数,保留系数较小的两类标准样本,然后采用Hausdorff距离算法计算待识别车辆轮廓与剩余两类样本之间的匹配程度,认定Hausdorff距离较小的两者具有相同的车型.实验结果表明,该方法准确、有效且实时性较好,在高速公路收费站、自动收费停车场等场合具有较大的实用价值.

An automotive-type recognition method based on the orientation and matching of contour

A new automotive-type recognition method based on the orientation and matching of contour is proposed.This method first obtains the contour of moving vehicles based on background subtract,lo- cates the contour by external rectangle,selects the side-contours of saloon car,passenger car and truck as three standard templets,calculates the matching coefficients between the side-contour which need to be recognized and three standard templets reserves two standard templets with small coefficients,then calculates Hausdorff distance between the contour which need to be recognized and the reserved two templets.The small Hausdorff distance could be judged as the same automotive-type.Experimental re- sults show that the method has high recognition precision,better time-consuming and practical applied value in highway toll-station and automatic toll-park.