摘要
为解决城市道路日益严峻的拥堵问题,结合深度学习和图像处理技术,提出了一种基于卷积神经网络的道路拥堵检测方法.此方法相对于传统机器视觉方法,无需前期提取道路背景,不受光照亮度和实际环境的影响,具有识别速度快、占用计算资源少、泛化性好等特点.现已在实际项目中得以应用,并取得了较好的效果.
In order to solve the increasingly serious congestion problem of urban roads,a road congestion detection method based on convolutional neural network was proposed in this paper.This method was based on the application value of road congestion rapid detection technology to alleviate the traffic jam problem.Besides,it combined deep learning with image processing technology.Compared with traditional methods,this method did not need to extract the road background in the early stage,and was not affected by the illumination brightness and the actual environment.It had the characteristics of fast recognition speed,less occupied computing resources and good generalization.It has been applied in practical projects and achieved good results.
作者
罗荣辉
袁航
钟发海
聂上上
LUO Ronghui;YUAN Hang;ZHONG Fahai;NIE Shangshang(School of Physics Engineering,Zhengzhou University,Zhengzhou 450001,China)
出处
《郑州大学学报(工学版)》
CAS
北大核心
2019年第2期18-22,共5页
Journal of Zhengzhou University(Engineering Science)
基金
国家自然科学基金资助项目(61601322)
河南省科技攻关计划(高新领域)项目(162102210018)
关键词
卷积神经网络
深度学习
图像识别
拥堵检测
智慧城市
convolutional neural network
deep learning
image recognition
congestion detection
smart city