Detection and Elimination of Dynamic Feature Points Based on YOLO and Geometric Constraints.
Saved in:
| Title: | Detection and Elimination of Dynamic Feature Points Based on YOLO and Geometric Constraints. |
|---|---|
| Authors: | Lu, Jiajia1 (AUTHOR), Wang, Xianwei2 (AUTHOR), Tang, Yue1 (AUTHOR) lj110428@163.com, Xi, Kan1 (AUTHOR), Shen, Yue2 (AUTHOR), Chen, Weichao3 (AUTHOR) |
| Source: | Arabian Journal for Science & Engineering (Springer Science & Business Media B.V. ). Jan2025, Vol. 50 Issue 2, p759-773. 15p. |
| Subjects: | Optical flow, Fix-point estimation, Artificial intelligence, Image processing, Algorithms |
| Abstract: | For the problem where dynamic objects in complex environments have a large impact on simultaneous localization and mapping (SLAM) pose estimation and mapping accuracy, you only look once (YOLO) detection method combining optical flow and geometric constraints is proposed to identify and eliminate dynamic feature points. Firstly, the latent dynamic feature points in the environment are detected based on YOLO, and the static and dynamic regions are divided according to motion consistency. Secondly, motion detection and tracking are carried out based on the optical flow method to preliminarily estimate the motion state, and then, re-judgment is performed in combination with geometric constraints to reduce tracking loss and precision reduction caused by false elimination of feature points. Finally, on the basis of eliminating dynamic feature points, static feature points are used for pose estimation to avoid the interference of dynamic feature points on pose estimation. Finally, based on the Technical University Munich (TUM) dataset and SLAM accuracy evaluation indexes, such as absolute trajectory error (ATE), and so on, the proposed method of this paper is experimentally tested and evaluated, and the ATE index of this paper's algorithm is improved by 8.1% and 96.35% compared with ORB-SLAM2 under TUM's low-dynamic sequences and high dynamic sequences, respectively, and the results show that this paper's algorithm has a better accuracy under the dynamic environment. [ABSTRACT FROM AUTHOR] |
| Copyright of Arabian Journal for Science & Engineering (Springer Science & Business Media B.V. ) is the property of Springer Nature and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Database: | Engineering Source |
|
Full text is not displayed to guests.
Login for full access.
|
|
| Abstract: | For the problem where dynamic objects in complex environments have a large impact on simultaneous localization and mapping (SLAM) pose estimation and mapping accuracy, you only look once (YOLO) detection method combining optical flow and geometric constraints is proposed to identify and eliminate dynamic feature points. Firstly, the latent dynamic feature points in the environment are detected based on YOLO, and the static and dynamic regions are divided according to motion consistency. Secondly, motion detection and tracking are carried out based on the optical flow method to preliminarily estimate the motion state, and then, re-judgment is performed in combination with geometric constraints to reduce tracking loss and precision reduction caused by false elimination of feature points. Finally, on the basis of eliminating dynamic feature points, static feature points are used for pose estimation to avoid the interference of dynamic feature points on pose estimation. Finally, based on the Technical University Munich (TUM) dataset and SLAM accuracy evaluation indexes, such as absolute trajectory error (ATE), and so on, the proposed method of this paper is experimentally tested and evaluated, and the ATE index of this paper's algorithm is improved by 8.1% and 96.35% compared with ORB-SLAM2 under TUM's low-dynamic sequences and high dynamic sequences, respectively, and the results show that this paper's algorithm has a better accuracy under the dynamic environment. [ABSTRACT FROM AUTHOR] |
|---|---|
| ISSN: | 2193567X |
| DOI: | 10.1007/s13369-024-08957-z |
