Based on the requirements of intelligent transportation system (ITS) , an algorithm for front-vehicle detection is proposed in this paper. With the algorithm, firstly a Haar-like feature cascade classifier is designed by using feature detection method to effectively identify the rear features of front vehicles. Then aiming at the problems of duplicate detection, noises, undections and the failure in detecting distant objects due to the limitation in resolution ratio, which occurred in real detections with complicated road conditions, the algorithm is revised. Finally the experiments and analyses are performed again with a result showing that the revised algorithm can effectively detect the front vehicles realtime in complex background by using on-board video camera, being easy to apply to ITS.%本文中根据智能交通系统对于前方车辆识别的需求,提出一种前车识别算法.首先利用特征检测法设计了基于Haar特征的车辆级联分类器,对车辆的尾部特征进行有效识别；然后针对真实复杂道路测试中出现的重复识别、噪声、漏检和受限于分辨率而难以识别远距离目标等问题,改进了算法；最后再次进行了实验和分析.结果表明,改进后的算法能有效利用车载摄像头实时识别复杂背景下的前方行驶车辆,便于在智能交通系统中应用.
展开▼
