基于深度学习的车联网图像识别系统设计  

Design of Image Recognition System for Internet of Vehicles Based on Deep Learning

作　　者：陈坚[1] 唐昌华[1] CHEN Jian;TANG Changhua(College of Humanities&Information Changchun University of Technology,Changchun Jilin 130122,China)

机构地区：[1]长春工业大学人文信息学院,吉林长春130122

出　　处：《信息与电脑》2022年第20期1-3,共3页Information & Computer

基　　金：吉林省教育厅科学技术研究项目“基于深度学习的图像超分辨率重构研究”(项目编号:JJKH20221281KJ)。

摘　　要：由于传统系统在车联网图像识别应用中的漏识率较高,基于深度学习设计车联网图像识别系统。硬件方面,设计了高清摄像机和发光二极管(Light Emitting Diode,LED)补光灯,获取道路车辆图像信息;软件方面,根据图像对比结果将车辆图像分成两类,对亮度较低的一类进行增强处理,利用深度学习网络模型对图像特征进行深度挖掘,并对提取的图像特征进行降维处理,根据图像特征与车联网数据库中的车辆图像进行对比,计算出图像相似度,识别到图像属性信息,从而实现基于深度学习的车联网图像识别系统设计。实验结果表明,设计系统对于车联网图像识别的漏识率低于传统系统,能够为图像识别提供精准的依据。Because of the high leak rate of the traditional system in the image recognition application of the Internet of Vehicles, the Internet of Vehicles image recognition system is designed based on deep learning. In terms of hardware, high-definition camera and Light Emitting Diode(LED) fill light are designed to obtain road vehicle image information. In terms of software, the vehicle images are divided into two categories according to the image comparison results. The one with lower brightness is enhanced, the image features are deeply mined using the depth learning network model, and the extracted image features are reduced in dimension. According to the comparison between the image features and the vehicle images in the Internet of Vehicles database, the image similarity is calculated, and the image attribute information is recognized, So as to realize the design of image recognition system of Internet of Vehicles based on deep learning. The experimental results show that the leakage rate of the designed system for the image recognition of the Internet of Vehicles is lower than that of the traditional system, which can provide an accurate basis for image recognition.

关 键 词：深度学习 车联网 图像识别 

分 类 号：U471.15[机械工程—车辆工程]