D-Fusion SLAM: A VSLAM system optimized by integrating depth information and RGBD feature points

This paper introduces D-Fusion SLAM, an advanced VSLAM system based on ORB-SLAM2. D-Fusion SLAM integrates depth information to enhance feature selection, aiming to improve accuracy and speed in visual SLAM applications. The system mainly comprises two modules: an image grid filtering module based o...

Full description

Saved in:
Bibliographic Details
Published inScientific Bulletin. Series C, Electrical Engineering and Computer Science no. 1; p. 121
Main Authors Zheng, Qingchun, Wu, Moudong, Zhu, Peihao, Yang, Bin, Li, Shubo
Format Journal Article
LanguageEnglish
Published Bucharest University Polytechnica of Bucharest 01.01.2025
Subjects
Image filters
Modules
Pose estimation
Online AccessGet full text
ISSN2286-3540

Cover

More Information
Summary:This paper introduces D-Fusion SLAM, an advanced VSLAM system based on ORB-SLAM2. D-Fusion SLAM integrates depth information to enhance feature selection, aiming to improve accuracy and speed in visual SLAM applications. The system mainly comprises two modules: an image grid filtering module based on grayscale information and a feature selection module integrating depth information. These modules effectively eliminate redundant and useless features, ensuring highquality features for pose estimation. We tested D-Fusion SLAM on the TUM dataset and compared it with ORB-SLAM2. Experimental results demonstrate that D-Fusion SLAM outperforms ORB-SLAM2, significantly enhancing the system's accuracy and robustness.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:2286-3540