基于Opencv的行驶车辆检测与计数系统  

Driving Vehicle Detection and Counting System Based on OpencvWEI Jinqiang

作　　者：魏金强 胡丹丹 李志宇 HU Dandan;LI Zhiyu(Zhejang Wenzhou Yongtaiwen Expressway Co.,Ltd.,Wenzhou,Zhejiang Province,325000 China)

机构地区：[1]浙江温州甬台温高速公路有限公司,浙江温州325000

出　　处：《汽车知识》2023年第4期139-141,共3页Auto Know

摘　　要：针对实时获取交通流量的迫切需求,该文基于Opencv提出了一种行驶车辆检测与计数系统。首先,采用像素加权平均值算法以及高斯滤波算法实现监控视频流灰度化及平滑处理,可显著提高检测算法实时性并为后续车辆检测提供高对比度图像;随后提出了一种基于改进的行驶车辆提取算法,在传统的基于高斯混合模型的背景分割算法的基础上引入形态学处理,可以修复监控视频远视场内车辆轮廓,大大地提高了车辆检测准确率;最后考虑到算法本身固有的检测时耗,提出了一种基于检测框中心点的区间计数算法,相比于单一的虚拟检测线而言,从很大程度上减低了车流量漏检率。Aiming at the urgent need of real-time traffic flow acquisition,this paper proposes a driving vehicle detection and counting system based on Opencv.Firstly,pixel weighted average algorithm and Gaussian filtering algorithm are used to achieve gray-scale and smooth processing of monitoring video stream,which can significantly improve the real-time performance of detection algorithm and provide high contrast images for subsequent vehicle detection.Then,this paper proposed an improved moving vehicle extraction algorithm,which introduced morphological processing on the basis of the traditional background segmentation algorithm based on Gaussian mixture model,and could repair the vehicle contour in the hyperopic field of surveillance video,greatly improving the vehicle detection accuracy.Finally,considering the inherent detection time consumption of the algorithm itself,an interval counting algorithm based on the center point of the detection frame is proposed.Compared with a single virtual detection line,the missing rate of traffic flow is reduced to a largeextent.

关 键 词：车辆检测 车辆计数 OPENCV 机器视觉 

分 类 号：U461[机械工程—车辆工程]