基于卷积神经网络的高速公路风险环境识别研究  

Research on Highway Risk Environment Identification Based on Convolutional Neural Network

作　　者：彭波 Peng Bo(Ningxia Jiaotou Expressway Management Co.,Ltd.,Yinchuan,China)

机构地区：[1]宁夏交投高速公路管理有限公司,宁夏银川

出　　处：《科学技术创新》2023年第14期117-120,共4页Scientific and Technological Innovation

基　　金：宁夏回族自治区交通科技项目-宁夏高速公路网典型风险路段交通安全保障提升关键技术研究(202100182)。

摘　　要：本研究基于学习算法对高速公路团雾天气图像进行识别。首先,对高速公路图像进行二分类识别,识别团雾图像与无团雾图像,确定效果良好的算法;其次,进一步对高速公路团雾图像根据能见度进行四分类,对高速公路团雾图像进行等级划分识别。发现8层卷积神经网络具有良好的识别效果,可用于高速公路团雾预警诱导系统,提升高速公路管理水平和安全水平。In this research,a learning algorithm is used to identify the fog weather image of expressway.Firstly,the binary recognition is carried out to identify the image of expressway,and the algorithm with good effect is determined.Secondly,the highway fog image is further classified into four categories according to the visibility,and the highway fog image is classified and recognized.It is found that the eight-layer convolutional neural network has a good recognition effect and can be used in the expressway fog warning and induction system to improve the expressway management level and safety level.

关 键 词：交通安全 高速公路 团雾 图像识别 卷积神经网络 

分 类 号：U495[交通运输工程—交通运输规划与管理]