【摘要】 介绍了一种用固定的单摄像头拍摄交通图像,并从图像序列中检测、跟踪、分类车辆的方法。该方法大致可分为3部分:抽取背景图像和图像分割;基于针孔模型的摄像机定标,计算透视投影矩阵;利用区域特征进行匹配跟踪,建立目标链,恢复目标三维信息,采用模型匹配法对车型分类。实验证明方法是简单可行的。更多还原

【Abstract】 This paper introduces a method of detection,tracking and classification of vehicles in monocular image sequences of traffic scenes recorded by a stationary camera. This method is composed of three main steps. Firstly, background image can be extracted from the sequences and using which we can segment objects from input images. Secondly, perspective projection matrix is computed with camera calibration based on pin-hole model. Finally,region charactersare used for matched tracking, the establishment of object linkage and the recovery of 3-D information of the object. Model matching method is used for the classification of viehicles. Experiments show that the method is simple and feasible.更多还原