Visual odometry based on camera motion and bidirectional long short-term memory network

Zhangzhen Zhao; Bin Xing; Tao Song; Yi Liu; Bin Chen

doi:10.1117/12.2656772

20 October 2022 Visual odometry based on camera motion and bidirectional long short-term memory network

Zhangzhen Zhao, Bin Xing, Tao Song, Yi Liu, Bin Chen

Proceedings Volume 12451, 5th International Conference on Computer Information Science and Application Technology (CISAT 2022); 124514N (2022) https://doi.org/10.1117/12.2656772
Event: 5th International Conference on Computer Information Science and Application Technology (CISAT 2022), 2022, Chongqing, China

Abstract

Most of the deep learning-based visual odometers have defects such as being unable to effectively learn camera motion information directly from RGB images, and low pose accuracy in the face of vehicle turning, high-speed driving and other scenes with dramatic changes in motion. Aiming at the above problems, this paper proposes a monocular visual odometry method that combines the camera's own motion estimation and a bidirectional long-short-term memory network. Firstly, a camera motion based on a gated recurrent convolutional neural network is proposed by introducing a 4D correlation volume of pixel pairs. Feature extraction algorithm to extract the motion features of adjacent frames of the camera; Then, a monocular visual odometry combining motion features and bidirectional long-short-term memory network is proposed. By learning the geometric motion relationship between adjacent frames before and after, more accurate six-degree-of-freedom camera pose information can be obtained. Comparative experiments on the KITTI dataset show that the proposed algorithm can estimate the pose with higher accuracy.

Citation Download Citation

Zhangzhen Zhao, Bin Xing, Tao Song, Yi Liu, and Bin Chen "Visual odometry based on camera motion and bidirectional long short-term memory network", Proc. SPIE 12451, 5th International Conference on Computer Information Science and Application Technology (CISAT 2022), 124514N (20 October 2022); https://doi.org/10.1117/12.2656772

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available

Members: $17.00

Non-members: $21.00 ADD TO CART

PROCEEDINGS
10 PAGES

DOWNLOAD PAPER SAVE TO MY LIBRARY

GET CITATION

RIGHTS & PERMISSIONS

Get copyright permission Get copyright permission on Copyright Marketplace

KEYWORDS

Cameras

Visualization

Optical flow

Feature extraction

Motion models

Convolutional neural networks

Error analysis

Show All Keywords

Keywords/Phrases

Search In:

Publication Years