毕业设计说明书（论文)中文摘要摘要：随着人们对出行要求的提高，交通安全也成为了人们最关心的问题之一，而交通标线的识别检测技术也是确保交通安全的关键问题。本文通过行车视频图像预处理技术，总结了传统基于霍夫变换实现的交通标线识别技术，并提出了改进优化方法。在此基础上，分析了传统检测算法具有的哪些弊端，提出基于卷积神经网络来实现交通标线的检测，探索了算法改进的优化方法，并将其应用在了实线违法变道的检测当中。关键词：行车视频；交通标线检测；卷积神经网络毕业设计说明书（论文）外文摘要Title Research on traffic marking recognition technologybased on driving videoAbstractWith the improvement of people's travel requirements,traffic safety hasbecome one of the most concerned issues,and the identification anddetection technology of traffic line marks is also the key issue to ensuretraffic safety.Based on the pre-processing technology of driving videoimage,this paper summarizes the traditional traffic line identificationtechnology based on Hough transform,and puts forward an improvedmethod.On this basis,the disadvantages of the traditional detectionalgorithm are analyzed,and the convolutional neural network is proposedto realize the detection of traffic marks.The optimization method of theimproved algorithm is explored,and it is applied to the detection ofillegal lane changes on solid lines.Keywords:Driving video;Traffic line marking detection;CNN本科毕业设计说明书（论文)目录1.1研究背景与意义11.2国内外研究现状...21.3研究内容与章节安排482.1感兴趣区域(R0I)提取..8113基于传统方法的交通标线识别..13133.2传统算法的检测和改进...154基于卷积神经网络的交通标线识别184.1卷积神经网络(CNN)....·184.2基于CNN的交通标线检测...20235交通标线检测实例应用及模型分析29295.2应用实例功能与总体设计..29325.4系统检测分析33结论353637