室外眩光场景ORB-SLAM2稳健定位模型研究

doi:10.13474/j.cnki.11-2246.2023.0265

测绘通报 ›› 2023, Vol. 0 ›› Issue (9): 59-63.doi: 10.13474/j.cnki.11-2246.2023.0265

室外眩光场景ORB-SLAM2稳健定位模型研究

吴思齐¹, 刘飞², 白羽¹, 马运涛², 王斐³, 郭梓钰²

1. 北京建筑大学理学院, 北京 102616;
2. 北京建筑大学测绘与城市空间信息学院, 北京 102616;
3. 东方通用航空摄影有限公司, 山西太原 030031

收稿日期:2022-11-14 发布日期:2023-10-08
通讯作者: 刘飞。E-mail:liufei1@bucea.edu.cn
作者简介:吴思齐(1999—),男,硕士生,主要从事视觉定位和深度学习领域的研究。E-mail:2102520021006@stu.bucea.edu.cn
基金资助:
国家自然科学青年基金(42104017);北京建筑大学双塔计划(JDYC20220825);北京建筑大学青年教师科研能力提升计划(X21021);北京建筑大学研究生创新项目(PG2022119);基于“北斗+加速度计”长城环境振动灾害监测技术研究(BGY2021KY-06Z)

Research on the robust positioning model of ORB-SLAM2 in outdoor glare scene

WU Siqi¹, LIU Fei², BAI Yu¹, MA Yuntao², WANG Fei³, GUO Ziyu²

1. School of Science, Beijing University of Civil Engineering and Architecture, Beijing 102616, China;
2. School of Geomatics and Urban Spatial Informatics, Beijing University of Civil Engineering and Architecture, Beijing 102616, China;
3. Eastern General Aerophotography Co., Ltd., Taiyuan 030031, China

Received:2022-11-14 Published:2023-10-08

摘要/Abstract

摘要： 为解决视觉SLAM在户外眩光场景下稳定性较低的问题,本文结合深度学习在图像处理领域的方法,提出了一种面向室外含有眩光干扰场景的改进的ORB-SLAM2算法。首先,将定位相机拍摄图片输入预先训练的U-Net网络处理,该网络可有效去除图片眩光,同时对无眩光干扰的图像可保持原始信息输出;然后,采用图像亮度判断算法判断该网络输出图像的亮度,对低亮度图像采用对比度受限的直方图均衡化处理(CLAHE)。根据实际户外场景下的试验结果,本文改进后的ORB-SLAM2系统的运动轨迹相比于原始系统,更加接近相机真实运动轨迹,同时该系统在户外眩光场景下具有更好的稳健性。

关键词: 视觉SLAM, 眩光, U-Net, 直方图均衡化处理, 稳健性

Abstract: In order to solve the problem of low stability of visual SLAM in outdoor glare scenes, combined with the method of deep learning in the field of image processing, an improved ORB-SLAM2 algorithm for outdoor scenes containing glare interference is proposed. Firstly, the images taken by the positioning camera are fed into the pre-trained U-Net network for processing, which can effectively remove the glare of the pictures, while maintaining the original information output for the images without glare interference. Secondly, the image brightness judgment algorithm used to judge the brightness of the output images of the network, and the contrast limited histogram equalization is used for the low-brightness images. According to the experimental results in actual outdoor scenes, the motion trajectory of the improved ORB-SLAM2 system is closer to the real motion trajectory of the camera than the original system, and the system has better robustness in outdoor glare scenes.

Key words: visual SLAM, glare, U-Net, CLAHE, robustness

中图分类号:

P228

吴思齐, 刘飞, 白羽, 马运涛, 王斐, 郭梓钰. 室外眩光场景ORB-SLAM2稳健定位模型研究[J]. 测绘通报, 2023, 0(9): 59-63.

WU Siqi, LIU Fei, BAI Yu, MA Yuntao, WANG Fei, GUO Ziyu. Research on the robust positioning model of ORB-SLAM2 in outdoor glare scene[J]. Bulletin of Surveying and Mapping, 2023, 0(9): 59-63.

参考文献

[1] 刘浩敏,章国锋,鲍虎军.基于单目视觉的同时定位与地图构建方法综述[J].计算机辅助设计与图形学学报,2016,28(6):855-868.
[2] BRESSON G,ALSAYED Z,YU L,et al.Simultaneous localizations and mapping:a survey of current trends in autonomous driving[J].IEEE Transactions on Intelligent Vehicles,2017,2(3):194-220.
[3] NEWCOMBE R A,LOVEGROVE S J,DAVISON A J.DTAM:dense tracking and mapping in real-time[C]//Proceedings of 2011 International Conference on Computer Vision.Los Alamitos:IEEE Computer Society Press,2011:2320-2327.
[4] WU X,PRADALIER C.Illumination robust monocular direct visual odometry for outdoor environment mapping[C]//Proceedings of 2019 International Conference on Robotics and Automation.[S.l.]:ICRA,2019:2392-2398.
[5] MUR-ARTAL R,MONTIEL J M M,TARDOS J D.ORB-SLAM:a versatile and accurate monocular SLAM system[EB/OL].[2022-09-20].https://arxiv.org/abs/1502.00956.
[6] 王林生.防眩光用水性聚氨酯涂层涂料的合成与性能研究[D].北京:北京化工大学,2013.
[7] ASHA C S,BHAT S K,NAYAK D,et al.Auto removal of bright spot from images captured against flashing light source[C]//Proceedings of 2019 IEEE International Conference on Distributed Computing,VLSI,Electrical Circuits and Robotics (DISCOVER).[S.l.]:IEEE,2019:1-6.
[8] CHABERT F.Automated lens flare removal[R].[S.l.]:Stanford University,Department of Electrical Engineering, 2015.
[9] VITORIA P,BALLESTER C.Automatic flare spot artifact detection and removal in photographs[J].Journal of Mathematical Imaging and Vision,2019,61(4):515-533.
[10] FAN Qingnan,YANG Jiaolong,HUA Gang,et al.A generic deep architecture for single image reflection removal and image smoothing[EB/OL].[2022-09-15].https://arxiv.org/abs/1708.03474.
[11] LI C,YANG Y X,HE K,et al.Single image reflection removal through cascaded refinement[C]//Proceedings of 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition.[S.l.]:IEEE,2020:3565-3574.
[12] ZHANG X E,NG R,CHE Q F.Single image reflection separation with perceptual losses[C]//Proceedings of 2018 IEEE Conference on Computer Vision and Pattern Recognition.[S.l.]:IEEE,2018:4786-4794.
[13] QIAN R,TAN R T,YANG W H,et al.Attentive generative adversarial network for raindrop removal from a single image[C]//Proceedings of 2018 IEEE Conference on Computer Vision and Pattern Recognition.[S.l.]:IEEE,2018:2482-2491.
[14] WEI W,MENG D Y,ZHAO Q,et al.Semi-supervised transfer learning for image rain removal[C]//Proceedings of 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition.[S.l.]:IEEE,2019:3877-3886.
[15] WU Y C,HE Q R,XUE T F,et al.How to Train Neural Networks for Flare Removal[C]//Proceedings of 2021 IEEE/CVF International Conference on Computer Vision.[S.l.]:IEEE,2021:2239-2247.
[16] DAI Y,LI C,ZHOU S,et al.Flare7K:a phenomenological nighttime flare removal dataset[EB/OL].[2022-09-30].https://doi.org/10.48550/arXiv.2210.06570.

室外眩光场景ORB-SLAM2稳健定位模型研究

Research on the robust positioning model of ORB-SLAM2 in outdoor glare scene

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

本文评价

[1]	赵丽凤, 王勇, 王晓静, 任传斌, 徐鹏宇. 改进U-Net模型支持下的高密度激光点云在沥青道路病害识别中的应用[J]. 测绘通报, 2023, 0(7): 136-141,159.
[2]	王本礼, 王也, 唐先龙, 董胜光. 遥感影像变化图斑智能化提取平台研发与应用[J]. 测绘通报, 2023, 0(2): 150-154.
[3]	冯炜明, 张新长, 孙颖, 姜明, 甘巧, 侯幸幸. 融合Transformer结构的高分辨率遥感影像变化检测网络[J]. 测绘通报, 2022, 0(8): 36-40,92.
[4]	窦世卿, 郑贺刚, 徐勇, 陈治宇, 苗林林, 宋莹莹. 基于U-Net3+的高分遥感影像建筑物提取[J]. 测绘通报, 2022, 0(6): 40-44.
[5]	高鸣, 周鑫鑫, 刘琦, 杨光迪, 吴长彬. 深度学习网络支持下的农房侵占耕地自动化监测[J]. 测绘通报, 2022, 0(3): 47-53.
[6]	普东东, 丁海勇. 融合超体素与成对链接聚类的LiDAR点云分割算法[J]. 测绘通报, 2022, 0(3): 65-69.
[7]	车子杰, 高飞, 吴兆福, 李振轩. 基于改进U-Net网络的多源遥感影像洪涝灾害信息提取与变化分析[J]. 测绘通报, 2022, 0(1): 26-32.
[8]	王学文, 赵庆展, 田文忠, 龙翔, 江萍. U-Net模型对不同空间分辨率防护林提取精度的影响[J]. 测绘通报, 2021, 0(6): 39-43.
[9]	王守志, 张福坤, 朱鹏飞, 詹昊, 张云姣, 奚歌. 材质和倾角属性筛选的光伏屋顶提取[J]. 测绘通报, 2021, 0(3): 96-99,151.
[10]	黄磊, 陈尔东昊. 基于多特征U-Net网络的工程活动图斑快速识别[J]. 测绘通报, 2021, 0(12): 75-78.
[11]	张永宏, 蔡朋艳, 陶润喆, 王剑庚, 田伟. 基于改进U-Net网络的遥感图像云检测[J]. 测绘通报, 2020, 0(3): 17-20,34.
[12]	盛超, 潘树国, 赵涛, 曽攀, 黄砺枭. 基于图像语义分割的动态场景下的单目SLAM算法[J]. 测绘通报, 2020, 0(1): 40-44.
[13]	许慧敏, 齐华, 南轲, 陈敏. 结合nDSM的高分辨率遥感影像深度学习分类方法[J]. 测绘通报, 2019, 0(8): 63-67.
[14]	张良桥, 陈国良, 许晓东, 连达军, 王睿. 一种用于图像特征提取的改进ORB-SLAM算法[J]. 测绘通报, 2019, 0(3): 16-20.
[15]	陈向阳, 向云飞. 利用LMedS算法提取规则建筑物顶部面片[J]. 测绘通报, 2018, 0(4): 121-124.