基于转折角点的车辆遮挡分离算法  

Vehicle Occlusion Separation Algorithm based on Turn Angular Point

作　　者：贾斌[1] 张洪刚[1] 

机构地区：[1]北京邮电大学信息与通信工程学院模式识别实验室,北京市100876

出　　处：《软件》2012年第12期122-125,共4页Software

摘　　要：本文提出了一种基于转折角点的车辆遮挡分离算法,该算法分为目标区域块的空洞填充、转折角点的提取和分类、基于转折角点的模板匹配三个步骤。在目标区域块的空洞填充步骤中,本文使用基于轮廓的空洞填充方法;在转折角点的提取和分类步骤中,本文通过圆盘移动方法提取目标区域块的转折角点并通过转折角点的倾斜角归一化直方图对转折角点进行分类;在基于转折角点的模板匹配步骤中,本文利用目标区域块和模板两者对应的循环链表进行比对,找出与目标区域块匹配的模板,根据模板提供的分离策略,实现遮挡分离。本文提出的算法适用于各类常见的遮挡情况,具有广泛的适用性。This paper puts forward occlusion vehicle separation algorithm based on turn angular point. This algorithm can be divided into the interspace filling of the target area, the extraction and classification of turn angle point and the template matching based on the turn angle point.In the step of interspace filling, interspace filling algorithm based on contour is put forward. In the step of extraction and classification,disc movement algorithm and slant angle normalized histogram algorithm are put forward.In the step of template matching,this paper compares the data of occlusion vehicle with the template library and finds out the matching template.This occlusion separation algorithm put forward in this paper can be applied to all kinds of common occlusion, and the algorithm has good feasibility.

关 键 词：计算机图象处理 车辆检测 遮挡分离 转折角点 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]