Retinex-based sparse direct VSLAM 3D reconstruction

被引：0

作者：

Xia L. ^{[1
]}

Meng Y. ^{[1
]}

Wang C. ^{[1
]}

Liu R. ^{[1
]}

机构：

[1] School of Automation Engineering, Northeast Electric Power University, Jilin

来源：

| 1600年 / Editorial Department of Journal of Chinese Inertial Technology卷 / 29期

关键词：

3D reconstruction; Loop closure detection; RDSOL; Single scale Retinex; VSLAM;

D O I：

10.13695/j.cnki.12-1222/o3.2021.02.002

中图分类号：

学科分类号：

摘要：

Visual simultaneous localization and mapping (VSLAM) has been widely used in the indoor robot navigation tasks. For the needs of autonomous navigation and map reusability, the real-time 3D reconstruction needs to overcome the problems of uneven illumination, low quantity and quality of corner extraction, and pose drift, etc. Therefore, a class of frameworks of Retinex-based direct sparse odometry with loop closure detection (RDSOL) is built, which conforms to DSO prototype, while breaks the design in both front-end photometric calibration and non-loop closure detection of DSO in terms of algorithm. By means of single scale Retinex-based image enhancements (instead of photometric calibration) and an added loop closure, it is desired to address ORB feature missing, eliminate accumulated error of pose drifts and derive globally consistent trajectories and maps. The optimal 3D reconstruction by RDSOL under open-source dataset and real scene tests are all achieved as expected, and when compared to DSO on TUM-mono dataset, RDSOL presents a 10.91% runtime speedup, and the average time of the sequence is greatly reduced. © 2021, Editorial Department of Journal of Chinese Inertial Technology. All right reserved.

引用

页码：147 / 153

页数：6

共 11 条

[1]

Wu C., Towards linear-time incremental structure from motion, International Conference on 3D Vision IEEE Computer Society, pp. 127-134, (2013)

[2]

Sibley G, Mei C, Reid I, Et al., Vast-scale outdoor navigation using adaptive relative bundle adjustment, International Journal of Robotics Research, 29, 8, pp. 958-980, (2010)

[3]

Han L, Wang J., Research on global threshold segmentation method of color image based on two-dimensional histogram, Journal of Northeast Electric Power University, 40, 5, pp. 76-83, (2020)

[4]

Engel J, Usenko V, Cremers D., A photometrically calibrated benchmark for monocular visual odometry, (2016)

[5]

Park C, Song J., Illumination change compensation and extraction of corner feature orientation for upward-looking camera-based SLAM, International Conference on Ubiquitous Robots & Ambient Intelligence, pp. 224-227, (2015)

[6]

Ding X, Wang Y, Xiong R, Et al., Persistent stereo visual localization on cross-modal invariant map, IEEE Transactions on Intelligent Transportation Systems, 21, 11, pp. 4646-4658, (2020)

[7]

Zheng B, Liu Q, Zhao F, Et al., Loop detection and semantic mapping algorithm fused with semantic information, Journal of Chinese Inertial Technology, 28, 5, pp. 629-637, (2020)

[8]

Bergmann P, Wang R, Cremers D., Online photometric calibration of auto exposure video for realtime visual odometry and SLAM, IEEE Robotics & Automation Letters, 3, 2, pp. 627-634, (2018)

[9]

Deng G, Jia Y, Liu Y., Method for extracting region of interest in single subject image based on color features, Journal of Northeast Electric Power University, 40, 4, pp. 89-94, (2020)

[10]

Ma W, Liu J, Wang X, Et al., Adaptive image defogging algorithm combined with lab space and single scale Retinex, Journal of Applied Optics, 41, 1, pp. 100-106, (2020)

← 1 2 →