基于视频流的多运动车辆检测算法研究  

Research on Multiple Moving Vehicles Detection Algorithm Based on Video Stream

作　　者：徐志佳[1] 黄昊旻[1] 舒泽芳[1] XU Zhi-jia;HUANG Hao-min;SHU Ze-fang(College of Mechanical Engineering,Guiyang University,Guiyang 550005 Guizhou,China)

机构地区：[1]贵阳学院机械工程学院,贵州贵阳550005

出　　处：《贵阳学院学报（自然科学版）》2018年第2期86-88,共3页Journal of Guiyang University：Natural Sciences

基　　金：贵州省科学技术基金项目:"视频流多车牌识别系统的关键技术研究"(项目编号:黔科合LH字[2014]7206)

摘　　要：本文提出了一种优化的视频流多运动车辆检测算法,该方法克服了传统的背景差法对阈值选取的依赖性,通过混合高斯背景建模获取前景图,并结合多连通区域的轮廓提取对多运动目标车辆检测。基于MATLAB R2016a平台进行实验,多运动车辆检测效果较好,验证了该算法的可行性。An optimization multiple moving vehicles detection algorithm based on video stream is proposed. This method overcomes the dependence of traditional background difference on the threshold selection. Extract foreground image by the gaussian mixture modeling. Afterwards, eontonr extration of multi - connected region is adopted to detect multiple moving objective vehicles. The feasibility of the proposed algorithm is proved by the experiment based on MATLAB R2016a palt-form. Experimental results show that multiple moving vehicles detection achieve good results.

关 键 词：多运动车辆检测 混合高斯模型 多连通区域 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]