| 265 | 5 | 7 |
| 下载次数 | 被引频次 | 阅读次数 |
夜间等低照度环境下，光源复杂、采集图像伴有噪声，图像细节信息弱化，造成夜间车辆目标特征提取困难，误检率、漏检率高。本文以夜间微弱路灯下车辆为研究目标，采集低照度环境中车辆图片，构造夜间车辆数据集，对低照度车辆数据集中部分较暗图像进行低照度图像增强处理，并使用增强后的图片对夜间车辆数据集进行扩展。采用YOLOv5s建立夜间车辆检测模型，并在此基础上采用ShuffleNetv2的思想对网络继续优化，结构改进后的检测网络对夜间远处车辆和被遮挡车辆的检测效果更好。
Abstract：At night or in other low illumination environments, the light source is complex, the acquisition image is accompanied by noise, and the detail information of the image is weakened, which result in the difficulty of vehicle target feature extraction, high false detection rate and high missed rate at night. The vehicle images under weak street lights at night is taken as the research target. The vehicle data sets in low illumination environment are constructed, and part of the darker images of the data sets are enhanced. Next, the enhanced image is paired with the original image as a training set and input into YOLOv5 to establish the vehicle detection model at night,and ShuffleNetv2 is adopted to further optimize the network. The improved detection network performs better on detecting distant vehicles and blocked vehicles at night.
[1]H. Kuang,X. Zhang,Y. Li,et al. Nighttime Vehicle Detection Based on Bio-Inspired Image Enhancement and Weighted Score-Level Feature Fusion[J]. IEEE Transactions on Intelligent Transportation Systems,2017,18(4):927-936.
[2]周敏飞.低照度图像增强与车辆识别技术研究[D].上海：东华大学，2020.
[3]曾锦华，邱秀莲，卞新伟，等.图像降噪和增强对人脸识别系统识别性能的影响[J].刑事技术，2021,46(1):6.
[4]董雪雪.夜间车辆检测算法研究及其应用[D].江苏：南京信息工程大学，2017.
[5]Girshick R, Donahue J, Darrell T, et al. Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation[C]. IEEE Computer Society,2013.
[6]Redmon J, Divvala S, Girshick R, et al. You Only Look Once:Unified, Real-Time Object Detection[C]. Computer Vision&Pattern Recognition. IEEE,2016.
[7]Redmon J, Farhadi A. YOLO9000:Better, Faster, Stronger[C].IEEE Conference on Computer Vision&Pattern Recognition. IEEE,2017:6517-6525.
[8]Redmon,J.,Farhadi,A. YOLOv3:An Incremental Improvement[J]. IEEE Trans. Pattern Anal.,2018(15):1125-1131.
[9]邓人玮，熊瑞平，卢文翔，等.基于YOLO神经网络模型的自动喷涂线工件分类识别方法[J].机械，2022,49(3):65-73.
[10]Bochkovskiy,A.,Wang,C. Y.,Liao,H. Y. M. YOLOv4:Optimal Speed and Accuracy of Object Detection[J]. arXiv 2020:arXiv.2004. 10934.
[11]Ultralytics. YOLOv5 2020[EB/OL]. https：//github. com/ultralytics/yolov5.
[12]Ma N,Zhang X,Zheng H T,et al. ShuffleNet V2:Practical Guidelines for Efficient CNN Architecture Design[J]. European Conference on Computer Vision(ECCV),2018:116-131.
[13]Loh Y P, Chan C S. Getting to know low-light images with the Exclusively Dark dataset[C]. Computer Vision and Image Understanding,2019.
[14]Wen L,Du D,Cai Z,et al. UA-DETRAC:A New Benchmark and Protocol for Multi-Object Detection and Tracking[C]. Computer Science,2015.
[15]Dong Z,Pei M,He Y,et al. Vehicle Type Classification Using Unsupervised Convolutional Neural Network[C]. IEEE transactions on intelligent transportation systems,2014.
[16]JAIN A K. Fundamentals of digital image processing[M]. New Jersey:Prentice-Hall, Inc.,1989.
[17]YING Zhen qiang,LI Ge,REN Yu rui,et al. A new low-light image enhancement algorithm using camera response model[C]. 2017IEEE International Conference on Computer Vision Workshop.Venice,Italy:IEEE,2017:3015.
基本信息:
DOI：
中图分类号:U495;TP391.41
引用信息:
[1]邹莹,龙伟,李炎炎等.低照度环境的车辆检测算法研究[J].机械,2022,49(07):66-74.
基金信息:
2021四川大学-遂宁市校地合作科技项目（2021CDSN-12）