基于图像处理的车距测量算法  

Distance Measurement Algorithm Based on Image Processing

作　　者：史战果[1,2] 张筱[1] 吴迪[1] 程帅[1] 张秀彬[1] 

机构地区：[1]上海交通大学电子信息与电气工程学院,上海200240 [2]南京军区兵种训练基地,江苏南京212400

出　　处：《电气自动化》2011年第5期78-80,共3页Electrical Automation

摘　　要：针对公路上行驶的汽车,驾驶员由于种种原因很难判定车距而造成汽车相撞的问题,提出了一种基于图像处理的车距测量算法。运用双目视感系统采集车辆前方实时图像,通过模板匹配对采集到的图像进行前方车辆识别,在确认有车辆后运用视差原理进行车距计算。实验表明,该算法运算周期小于30ms,识别准确率高达99%,在200m视距内距离检测误差小于1m。The difficulty of determining the distance between two cars for the drivers causes vehicle collision, a method of distance measurement algorithm based on image processing is presented. Acquire Real-time image in front of the vehicle by using binocular visual system, and determine whether vehicle exists in front through Template matching the image, then use the Parallax principle calculates the distance after confirming the existence of vehicle. It has been proved that the operation period of algorithm was less than 30 ms, the identification accuracy rate was more than 99%, and the tolerance of distance measurement was less than one meter on the condition that the range of visibility was 200 meters.

关 键 词：图像处理 模板匹配 视差原理 车距测量 运算周期 识别准确率 

分 类 号：TP202[自动化与计算机技术—检测技术与自动化装置]