毕业论文--基于视频图像的车道线识别

《毕业论文--基于视频图像的车道线识别》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、HUNANUNIVERSITY毕业论文论文题目基于视频图像的车道线识别学生姓名学生学号专业班级自动化4班学院名称电气与信息工程学院指导老师学院院长2015年5月25日摘要近年来，我国经济快速发展，工业水平得到了极大的提高，人们对汽车的需求量越来越大，然而交通事故日益严重，迫切需要对智能交通系统进行研究，车道线识别作为智能交通系统的基础技术之一，不可或缺。本文车道线识别的主要设计思路是：首先对图像进行缩放以及灰度化、平滑滤波处理、二值化处理、边缘增强和形态学操作，接着对得到的二值图像进行感兴趣区域的划分，得到车道线特征区域并进行筛选，经过

2、筛选后的二值图像，仅包含有两条车道线，所含数据量少、特征明显，所以对其进行Hough变换，检测出直线，最后通过最小二乘法对Hough变换得到的直线进行拟合。实验结果表明，本文设计的车道线识别算法能够较好的识别出视频图像包含的车道线信息，准确率较高，并且由于整个算法用的都是一些简单的图像处理操作，加上在筛选操作时仅保留两条车道线，所以具有较好的实时性。关键词：图像预处理，车道线特征区域提取，直线检测，车道线识别ILaneRecognitionBasedOnVideoImageAbstractWiththerapiddevelopmento

3、fnationaleconomyinrecentyears,thelevelofindustryhasbeengreatlyimproved.Trafficaccidentincreasinglyseriousbecauseofthegrowingdemandsforcar.Researchingintelligenttransportationsystemisimminent,andlanerecognitionasoneofthebasictechnologyforintelligenttransportationsystem,it

4、isindispensable.Thusthelanerecognitionhavebecomeanimportantresearchintheintelligenttransportationsystems.Themainprocessofthelanerecognitionsystemwasasfollows.First,processedtheimagewithaseriesofnecessarypretreatment,includingimagescaledandgrayscale,smoothingprocessing,bi

5、narization,edgeenhancementandmorphologicaloperations.Second,tookuseofthenaturalboundarybetweenroadregionandnon-roadregion,toextractlanefeatureregionbyscanningtheimagefrombottomtoupindichotomy.Thenbasedonthespecificcharacteristicsofmorphologicalanddistribution,dividethebi

6、naryimageonregionofinterestandscreenedthelanefeatureregionwhichgetjust.Thisprocessmakethebinaryimageonlycontaintwolaneslines,whichcontainslessdataandobviouscharacteristics.SoweuseHoughtransformtodetectstraightlineandfinallythroughtheleastsquaresmethodtostraightlinefittin

7、g.Experimentalresultsshowthatthelanerecognitionalgorithmdesignbasedonvideoimagecanbetteridentifythelaneinformationwhichcontainedinthevideoimage.Whilethislanerecognitionalgorithmhashighaccuracy.Andbecauseofsomesimpleimageprocessingoperationsusedinthealgorithmcoupledwithsc

8、reeningoperationonlyretaintwolane,thelanerecognitionsystemenjoybetterreal-time.Keywords:Imagepreprocess