改进特征匹配的ORB-SLAM稠密建图算法

doi:10.19678/j.issn.1000-3428.0065825

摘要/Abstract

摘要：

对于同步定位与建图（SLAM）中主流的特征点法，特征匹配是估计相机运动的关键，然而在特征匹配过程中存在图像特征的局部特性、误匹配等问题，成为视觉SLAM的瓶颈。此外，特征点法生成的稀疏地图只能用于定位，无法满足更高层次的需求。针对ORB-SLAM3中ORB特征点匹配效率低且未能生成稠密地图的问题，提出一种改进的ORB-GMS匹配策略并在ORB-SLAM3系统中加入稠密点云构建线程来实现稠密建图。将运动平滑性约束作为特征点运动统计的方法，通过比较特征点邻域内的匹配数量和阈值快速判断当前匹配是否正确，将图片网格化并快速计算网格内特征点的匹配数量，进行相机的位姿估计。根据关键帧与相应位姿构建稠密点云地图，采用外点去除滤波和体素网格滤波减小点云规模。在TUM的RGB-D数据集上的实验结果表明，与ORB-SLAM3相比，该算法可以减少约50%的匹配耗时，同时在匹配数量上平均提升60%，定位平均误差降低32%。此外，与稀疏地图相比，该方法生成易于2次加工的稠密点云地图，扩大算法的应用场景。

关键词: 同步定位与建图, 特征点, 特征匹配, 基于网格的运动统计, 稠密建图, 点云滤波

Abstract:

In the mainstream feature-based Simultaneous Localization and Mapping(SLAM) method, feature matching is a key step in estimating camera motion. However, the local characteristics of image features cause widespread mismatch and have become a major bottleneck in visual SLAM. In addition, the sparse maps generated by the feature-based method can only be used for localization, as they do not satisfy higher-level requirements. To address the problems of low efficiency in ORB feature point matching and failure to generate dense maps in ORB-SLAM3, an improved ORB Grid-based Motion Statistics(ORB-GMS) matching strategy is proposed, whereby a dense point cloud construction thread is added to ORB-SLAM3 to realize dense mapping. The motion smoothness constraint is used for the feature point motion statistics method, and the number of matches in the feature point neighborhood and threshold are compared to efficiently determine whether the current match is correct. The gridded images are used for fast computation to perform camera pose estimation. Finally, the dense point cloud map is constructed according to the key frame and the corresponding pose, using the outlier point removal and voxel-grid filters to reduce the size of the point cloud. The experimental results on the RGB-D dataset of TUM show that compared with ORB-SLAM3, the proposed algorithm can reduce matching time by approximately 50% and average positioning error by 32%, while increasing the number of matches by an average of 60%. In addition, compared to sparse maps, this method generates dense point cloud maps that are easy for secondary processing, thereby expanding the application scenarios of the algorithm.

Key words: Simultaneous Localization and Mapping(SLAM), feature point, feature matching, Grid-based Motion Statistics(GMS), dense mapping, point cloud filter

刘洋, 陈俊, 胡诗佳, 赖佳华. 改进特征匹配的ORB-SLAM稠密建图算法[J]. 计算机工程, 2023, 49(10): 247-254.

Yang LIU, Jun CHEN, Shijia HU, Jiahua LAI. Improved Feature Matching ORB-SLAM Algorithm for Dense Mapping[J]. Computer Engineering, 2023, 49(10): 247-254.

https://www.ecice06.com/CN/Y2023/V49/I10/247

图/表 10

图1 OpenCV与ORB-SLAM3提取ORB特征点对比

Fig.1 Comparison between OpenCV and ORB-SLAM3 in extracting ORB feature points

图2 特征点对的运动核

Fig.2 Motion kernel of feature point pairs

图3 正确匹配和错误匹配支持度对应的概率分布

Fig.3 Probability distribution corresponding to support for correct matching and wrong matching support

图4 本文算法架构

Fig.4 Architecture of the proposed algorithm

图5 ORB-SLAM3、ORB-GMS与本文算法筛除误匹配的结果

Fig.5 The results of ORB-SLAM3, ORB-GMS and the propoesd algorithm for screening out mismatches

图6 稀疏地图以及滤波前后的稠密点云地图效果

Fig.6 Sparse map and dense point cloud map without filtering and after filtering

参考文献 27

1	MUR-ARTAL R , MONTIEL J M M , TARDOS J D . ORB-SLAM: a versatile and accurate monocular SLAM system. IEEE Transactions on Robotics: A publication of the IEEE Robotics and Automation Society, 2015, 31 (5): 1147- 1163.
2	MUR-ARTAL R , TARDÓS J D . ORB-SLAM2:an open-source SLAM system for monocular, stereo, and RGB-D cameras. IEEE Transactions on Robotics, 2017, 33 (5): 1255- 1262. doi: 10.1109/TRO.2017.2705103
3	CAMPOS C , ELVIRA R , RODRÍGUEZ J J G , et al. ORB-SLAM3:an accurate open-source library for visual, visual-inertial, and multimap SLAM. IEEE Transactions on Robotics, 2021, 37 (6): 1874- 1890. doi: 10.1109/TRO.2021.3075644
4	高翔, 张涛, 刘毅. 视觉SLAM十四讲: 从理论到实践. 北京: 电子工业出版社, 2017.
	GAO X , ZHANG T , LIU Y . Fourteen lectures on visual SLAM: from theory to practice. Beijing: Publishing House of Electronics Industry, 2017.
5	FU Q , YU H S , WANG X L , et al. Fast ORB-SLAM without keypoint descriptors. IEEE Transactions on Image Processing, 2022, 31, 1433- 1446. doi: 10.1109/TIP.2021.3136710
6	LIU H M , ZHANG Q Q , FAN B , et al. Features combined binary descriptor based on voted ring-sampling pattern. IEEE Transactions on Circuits and Systems for Video Technology, 2020, 30 (10): 3675- 3687. doi: 10.1109/TCSVT.2019.2943595
7	TANG J X , ERICSON L , FOLKESSON J , et al. GCNv2:efficient correspondence prediction for real-time SLAM. IEEE Robotics and Automation Letters, 2019, 4 (4): 3505- 3512.
8	DUAN Y Q, LU J W, WANG Z W, et al. Learning deep binary descriptor with multi-quantization[C]//Proceedings of Conference on Computer Vision and Pattern Recognition. Washington D.C., USA: IEEE Press, 2017: 4857-4866.
9	LIN K , LU J W , CHEN C S , et al. Unsupervised deep learning of compact binary descriptors. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2019, 41 (6): 1501- 1514. doi: 10.1109/TPAMI.2018.2833865
10	DAI Z, HUANG X H, CHEN W N, et al. A comparison of CNN-based and hand-crafted keypoint descriptors[C]//Proceedings of International Conference on Robotics and Automation. Washington D.C., USA: IEEE Press, 2019: 2399-2404.
11	ZUO X X, XIE X J, LIU Y, et al. Robust visual SLAM with point and line features[C]//Proceedings of International Conference on Intelligent Robots and Systems. Washington D.C., USA: IEEE Press, 2017: 1775-1782.
12	GOMEZ-OJEDA R , MORENO F A , ZUÑIGA-NOËL D , et al. PL-SLAM: a stereo SLAM system through the combination of points and line segments. IEEE Transactions on Robotics, 2019, 35 (3): 734- 746. doi: 10.1109/TRO.2019.2899783
13	LI Q A , KANG J A , WANG Y X , et al. An improved feature matching ORB-SLAM algorithm. Journal of Physics: Conference Series, 2020, 1693 (1): 1- 10.
14	杨弘凡, 李航, 陈凯阳, 等. 基于改进ORB算法的图像特征点提取与匹配方法. 图学学报, 2020, 41 (4): 548- 555. URL
	YANG H F , LI H , CHEN K Y , et al. Image feature points extraction and matching method based on improved ORB algorithm. Journal of Graphics, 2020, 41 (4): 548- 555. URL
15	廖泓真, 王亮, 孙宏伟, 等. 一种改进的ORB特征匹配算法. 北京航空航天大学学报, 2021, 47 (10): 2149- 2154. URL
	LIAO H Z , WANG L , SUN H W , et al. An improved ORB feature matching algorithm. Journal of Beijing University of Aeronautics and Astronautics, 2021, 47 (10): 2149- 2154. URL
16	TATENO K, TOMBARI F, LAINA I, et al. CNN-SLAM: real-time dense monocular SLAM with learned depth prediction[C]//Proceedings of Conference on Computer Vision and Pattern Recognition. Washington D.C., USA: IEEE Press, 2017: 6565-6574.
17	MATSUKI H , SCONA R , CZARNOWSKI J , et al. CodeMapping: real-time dense mapping for sparse SLAM using compact scene representations. IEEE Robotics and Automation Letters, 2021, 6 (4): 7105- 7112. doi: 10.1109/LRA.2021.3097258
18	ALEXA M , BEHR J , COHEN-OR D , et al. Computing and rendering point set surfaces. IEEE Transactions on Visualization and Computer Graphics, 2003, 9 (1): 3- 15. doi: 10.1109/TVCG.2003.1175093
19	MARTON Z C, RUSU R B, BEETZ M. On fast surface reconstruction methods for large and noisy point clouds[C]//Proceedings of International Conference on Robotics and Automation. Washington D.C., USA: IEEE Press, 2009: 3218-3223.
20	KAZHDAN M, BOLITHO M, HOPPE H. Poisson surface reconstruction[EB/OL]. [2022-08-15]. https://citeseerx.ist.psu.edu/viewdoc/download;jsessionid=1A7CA019302E9D7961804BC8EFE4604D?doi=10.1.1.153.4724&rep=rep1&type=pdf.
21	ROSTEN E, DRUMMOND T. Machine learning for high-speed corner detection[EB/OL]. [2022-08-15]. https://citeseerx.ist.psu.edu/viewdoc/download;jsessionid=1906A90E48189AEBC419B37BB20ACC55?doi=10.1.1.60.3991&rep=rep1&type=pdf.
22	LOWE D G . Distinctive image features from scale-invariant keypoints. International Journal of Computer Vision, 2004, 60 (2): 91- 110. doi: 10.1023/B:VISI.0000029664.99615.94
23	BAY H, TUYTELAARS T, VAN GOOL L. SURF: speeded up robust features[C]//Proceedings of European Conference on Computer Vision. Berlin, Germany: Springer, 2006: 404-417.
24	RUBLEE E, RABAUD V, KONOLIGE K, et al. ORB: an efficient alternative to SIFT or SURF[C]//Proceedings of International Conference on Computer Vision. Washington D.C., USA: IEEE Press, 2012: 2564-2571.
25	CALONDER M, LEPETIT V, STRECHA C, et al. BRIEF: binary robust independent elementary features[C]//Proceedings of European Conference on Computer Vision. Berlin, Germany: Springer, 2010: 778-792.
26	BIAN J W, LIN W Y, MATSUSHITA Y, et al. GMS: grid-based motion statistics for fast, ultra-robust feature correspondence[C]//Proceedings of Conference on Computer Vision and Pattern Recognition. Washington D.C., USA: IEEE Press, 2017: 2828-2837.
27	STURM J, ENGELHARD N, ENDRES F, et al. A benchmark for the evaluation of RGB-D SLAM systems[C]//Proceedings of International Conference on Intelligent Robots and Systems. Washington D.C., USA: IEEE Press, 2012: 573-580.

[1]	徐晓滨, 张云硕, 施凡, 常雷雷, 陶志刚. 基于特征匹配度与异类子模型融合的安全性评估方法[J]. 计算机工程, 2024, 50(8): 113-122.
[2]	叶智奇, 章国宝, 朱宏伟. 一种面向室内动态行人场景的激光SLAM算法[J]. 计算机工程, 2024, 50(6): 208-217.
[3]	杜晓英, 袁庆霓, 齐建友, 王晨, 杜飞龙, 任澳. 动态场景下基于语义分割的视觉SLAM方法[J]. 计算机工程, 2024, 50(3): 242-249.
[4]	徐浩宸, 刘满华. 基于多层次自注意力网络的人脸特征点检测[J]. 计算机工程, 2024, 50(2): 239-246.
[5]	徐春波, 闫娟, 杨慧斌, 王博, 吴晗. 基于目标检测和语义分割的视觉SLAM算法[J]. 计算机工程, 2023, 49(8): 199-206, 214.
[6]	崔云轩, 刘桂华, 余东应, 郭中远, 张文凯. 点线特征融合的激光雷达单目惯导SLAM系统[J]. 计算机工程, 2022, 48(7): 254-263.
[7]	刘鑫, 王忠, 秦明星. 多机器人协同SLAM技术研究进展[J]. 计算机工程, 2022, 48(5): 1-10.
[8]	房立金, 王科棋. 一种结合深度学习的运动重检测视觉SLAM算法[J]. 计算机工程, 2022, 48(5): 18-26.
[9]	余东应, 刘桂华, 曾维林, 冯波, 张文凯. 自适应窗隔匹配与深度学习相结合的RGB-D SLAM算法[J]. 计算机工程, 2021, 47(8): 224-233.
[10]	张海涛, 秦鹏程. 基于GMS与FPME的视频目标跟踪方法[J]. 计算机工程, 2021, 47(7): 226-231.
[11]	张宇, 胡立坤, 黄彬. 融合颜色与光照信息的改进AKAZE图像匹配算法[J]. 计算机工程, 2021, 47(6): 225-233.
[12]	傅由甲. 基于面部特征点的单幅图像人脸姿态估计方法[J]. 计算机工程, 2021, 47(4): 197-203,210.
[13]	田智慧, 马占宇, 魏海涛. 基于密度核心的出租车载客轨迹聚类算法[J]. 计算机工程, 2021, 47(2): 133-138.
[14]	马科伟, 张锲石, 康宇航, 任子良, 程俊. 移动机器人中视觉里程计技术综述[J]. 计算机工程, 2021, 47(11): 1-10.
[15]	康智慧, 王全玉, 王战军. 一种端到端的人脸对齐方法[J]. 计算机工程, 2021, 47(10): 207-213.

选择文件类型/文献管理软件名称

选择包含的内容