摘要
This paper proposes a stereo visual odometry algorithm, which can not only effectively run in a dynamic environment without being affected by dynamic objects, but also detect the position of dynamic objects. A mechanism of the quantitative histogram (QH) and the optical flow angle histogram (OFAH) of feature points is used to obtain dynamic points. Besides, a multi-feature fusion mechanism is proposed to perform binary segmentation and obtain the bounding box of dynamic object. In the experimental section, we evaluate our algorithm on the KITTI dataset and compare it with ORB-SLAM2. Experiments show that our method can not only improve the accuracy of pose estimation in a dynamic environment, but also detect the position of dynamic objects.
学科与关键词
- Optical Flow-Based Stereo Visual Odometry With Dynamic Object DetectionLiu, Yu*; Zhou, ZhiyuIEEE TRANSACTIONS ON COMPUTATIONAL SOCIAL SYSTEMS, 2022.被引 3
- SVD-SLAM: Stereo Visual SLAM Algorithm Based on Dynamic Feature Filtering for Autonomous DrivingTian, Liangyu; Yan, Yunbing*; Li, HaoranElectronics, 2023, 12(8): 1883.被引 2
- DynPL-SVO: A Robust Stereo Visual Odometry for Dynamic ScenesZhang, Baosheng; Ma, Xiaoguang*; Ma, Hong-Jun; Luo, ChunboIEEE Transactions on Instrumentation and Measurement, 2024, 73: 5006510.
