无人车行驶环境图像的几何测距*

doi:10.19562/j.chinasae.qcgc.2020.08.006

汽车工程 ›› 2020, Vol. 42 ›› Issue (8): 1034-1039.doi: 10.19562/j.chinasae.qcgc.2020.08.006

无人车行驶环境图像的几何测距^*

代金坤¹, 罗玉涛¹, 梁伟强²

1.华南理工大学机械与汽车工程学院,广州 510640;
2.广汽集团汽车工程研究院,广州 510640

收稿日期:2019-10-19 出版日期:2020-08-25 发布日期:2020-09-24
通讯作者: 罗玉涛,教授,工学博士,E-mail:ctytluo@scut.edu.cn。
基金资助:
*广东省科技计划项目(2016B010132001)资助。

Geometric Ranging of Unmanned Vehicle Driving Environment Image

Dai Jinkun¹, Luo Yutao¹, Liang Weiqiang²

1. School of Mechanical & Automotive Engineering, South China University of Technology, Guangzhou 510640;
2. GAC Automotive Engineering Institute, Guangzhou 510640

Received:2019-10-19 Online:2020-08-25 Published:2020-09-24

摘要/Abstract

摘要： 本文中提出了一种无人车行驶环境图像的几何测距方法。首先,利用迁移学习的方法改进Tiny-YOLOv2网络模型,对需识别的物体进行训练与检测,标记物体并定位物体在图像中的位置。其次,提出了一种通过物体分类、边缘检测及边缘拟合的方法,进一步提取物体的图像信息。最后,建立了一种基于空间几何理论的测距模型,结合物体先验信息实现了物体的距离测量。该方法在4 m以内88%以上的测量值误差不超过0.2 m,同时测量误差并没有随着距离的增加而有较大变化。

关键词: 几何测距, 单目视觉, 卷积神经网络, 迁移学习, 图像分割

Abstract: A geometric ranging method for the image of the driving environment of the unmanned vehicle is proposed. Firstly, the migration-learning method is used to improve the Tiny-YOLOv2 network model so as to train and detect the object to be identified, mark the object and locate the position of the object in the image. Secondly, a method of object classification, edge detection and edge fitting is proposed to further extract image information of the object. Finally, a ranging model based on spatial geometry theory is established, and the distance measurement of the object is realized by combining the prior information of the object size. With this method, more than 88% of the measurement error within 4 m is less than 0.2 m, and the measurement error does not change much with the increase of distance.

Key words: geometric ranging, monocular vision, convolutional neural network, migration learning, image segmentation

代金坤, 罗玉涛, 梁伟强. 无人车行驶环境图像的几何测距^*[J]. 汽车工程, 2020, 42(8): 1034-1039.

Dai Jinkun, Luo Yutao, Liang Weiqiang. Geometric Ranging of Unmanned Vehicle Driving Environment Image[J]. Automotive Engineering, 2020, 42(8): 1034-1039.

参考文献

[1] 樊露.单目摄像头精确测距技术及实现方法[J].装备制造技术,2016(3):185-187.
[2] VOGIATZIS G, CARLOS H. Video-based, real-time multi-view stereo[J]. Image & Vision Computing,2011,29(7):434-441.
[3] EIGEN D, FERGUS R. Predicting depth, surface normals and semantic labels with a common multi-scale convolutional architecture[C]. 2015 IEEE International Conference on Computer Vision (ICCV). IEEE,2015:2650-2658.
[4] TATENO K, TOMBAR F, LAINA I, et al. CNN-SLAM: real-time dense monocular SLAM with learned depth prediction[C]. 2017 IEEE Conference on Computer Vision and Pattern Recognition(CVPR)-CNN-SLAM,2017:6565-6574.
[5] ENGEL J, KOLTUN V, CREMERS D. Direct sparse odometry[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2017:611-625.
[6] 邹斌,袁宇翔.面向智能交通的单目视觉测距方法研究[J].交通运输系统工程与信息,2018(4):46-53.
[7] 李祎承,胡钊政,胡月志,等.基于GPS与图像融合的智能车辆高精度定位算法[J].交通运输系统工程与信息,2017,17(3):112-119.
[8] ROYER E, LHUILLIER M, DHOME M, et al. Monocular vision for mobile robot localization and autonomous navigation[J]. International Journal of Computer Vision,2007,74(3):237-260.
[9] 祝世平,强锡富.用于摄像机定位的单目视觉方法研究[J].光学学报,2001,21(3):339-343.
[10] 郭磊,徐友春,李克强,等.基于单目视觉的实时测距方法研究[J].中国图象图形学报,2018,11(1):74-81.
[11] 陈宗海,洪洋,王纪凯,等.基于循环卷积神经网络的单目视觉里程计[J].机器人,2019,41(2):13-21.
[12] YOUNES G, ASMAR D, SHAMMAS E, et al. Keyframe-based monocular SLAM: design, survey, and future directions[J]. Robotics and Autonomous Systems,2017,98:67-88.
[13] 李琳辉,伦智梅,连静,等.基于卷积神经网络的道路车辆检测方法[J].吉林大学学报(工学版),2017,47(2):384-391.
[14] GIRSHICK R, DONAHUE J, DARRELL T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]. Proceedings of 2014 IEEE International Conference on Computer Vision and Pattern Recognition. Columbus, OH: IEEE,2014:580-587.
[15] GIRSHICK R. Fast R-CNN[C]. Proceedings of 2015 IEEE International Conference on Computer Vision. Santiago: IEEE,2015:1440-1448.
[16] REN S, HE K, GIRSHICK R, et al. Faster R-CNN: towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence,2015,39(6):1137-1149.
[17] LIU W, ANGUELOV D, ERHAN D, et al. SSD: single shot multibox detector[C]. European Conference on Computer Vision. Springer, Cham, 2016:21-37.
[18] REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once: unified, real-time object detection[C].CVRR,2016:779-788.
[19] REDMON J, FARHADI A. YOLO9000: better, faster, stronger[C]. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR),2017:6517-6525.
[20] REDMON J, FARHADI A. YOLOv3: an incremental improvement[J/OL]. arXiv:1804.02767v1,2018.
[21] 徐德,谭民,李原.机器人视觉测量与控制[M].北京:国防工业出版社,2011.
[22] XU L, CAO Z, LIU X. A monocular vision system for pose measurement in indoor environment[C]. IEEE International Conference on Robotics & Biomimetics. IEEE,2017:1977-1982.

无人车行驶环境图像的几何测距^*

Geometric Ranging of Unmanned Vehicle Driving Environment Image

PDF (PC)

赞

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 12

Metrics

本文评价

推荐阅读 10

[1]	管欣,仲昭辉,詹军,奚腾龙,叶昊,高深圳,成健,廖世辉,蔡均. 基于卷积神经网络的汽车操纵稳定性试验类型识别方法[J]. 汽车工程, 2023, 45(9): 1765-1771.
[2]	王明,唐小林,杨凯,李国法,胡晓松. 考虑预测风险的自动驾驶车辆运动规划方法[J]. 汽车工程, 2023, 45(8): 1362-1372.
[3]	张雷, 关可人, 丁晓林, 郭鹏宇, 王震坡, 孙逢春. 基于图像识别与动力学融合的路面附着系数估计方法[J]. 汽车工程, 2023, 45(7): 1222-1234.
[4]	赵霞,李朝,付锐,葛振振,王畅. 基于深度卷积-Tokens降维优化视觉Transformer的分心驾驶行为实时检测[J]. 汽车工程, 2023, 45(6): 974-988.
[5]	汪梓豪,蔡英凤,王海,陈龙,熊晓夏. 基于单目视觉运动估计的周边多目标轨迹预测方法[J]. 汽车工程, 2022, 44(9): 1318-1326.
[6]	毕贵红,谢旭,蔡子龙,骆钊,陈臣鹏,赵鑫. 动态条件下基于深度学习的锂电池容量估计[J]. 汽车工程, 2022, 44(6): 868-878.
[7]	冯润泽,江昆,于伟光,杨殿阁. 基于两阶段分类算法的中国交通标志牌识别[J]. 汽车工程, 2022, 44(3): 434-441.
[8]	于海,邓钧君,王震坡,孙逢春. 基于卷积神经网络的逆变器故障诊断方法[J]. 汽车工程, 2022, 44(1): 142-152.
[9]	夏祥腾,王大方,曹江,赵刚,张京明. 基于稀疏卷积神经网络的车载激光雷达点云语义分割方法[J]. 汽车工程, 2022, 44(1): 26-35.
[10]	施冬梅,肖锋. 基于改进长短时记忆网络的驾驶行为检测方法研究[J]. 汽车工程, 2021, 43(8): 1203-1209.
[11]	王海, 王宽, 蔡英凤, 刘泽, 陈龙. 基于改进级联卷积神经网络的交通标志识别^*[J]. 汽车工程, 2020, 42(9): 1256-1262.
[12]	彭运赛, 夏飞, 袁博, 王志成, 罗志疆. 基于改进CNN和信息融合的动力电池组故障诊断方法^*[J]. 汽车工程, 2020, 42(11): 1529-1536.