A new visual odometry algorithm based on multi-path deep fully convolutional neural networks

Bo Chen; Kun Yan; Rongchuan Cao; Tianqi Zhang; Xiaoli Zhang

doi:10.1117/12.2653846

8 December 2022 A new visual odometry algorithm based on multi-path deep fully convolutional neural networks

Bo Chen, Kun Yan, Rongchuan Cao, Tianqi Zhang, Xiaoli Zhang

Proceedings Volume 12474, Second International Symposium on Computer Technology and Information Science (ISCTIS 2022); 124740W (2022) https://doi.org/10.1117/12.2653846
Event: Second International Symposium on Computer Technology and Information Science (ISCTIS 2022), 2022, Guilin, China

Abstract

Visual odometry is one of the key core technologies in the field of autonomous driving. However, images captured in lowlight or unevenly-illuminated scenes still cannot guarantee good performance due to low image contrast and lack of detail features. Therefore, we propose an end-to-end visual odometry method based on image fusion and FCNN-LSTM in the paper. The brightness image of the source image sequence is obtained by gray-scale transformation, and an image fusion algorithm based on spectral residual theory is designed to combine the image sequence and its brightness image to enhance the contrast of the image and provide more detailed information. In order to improve the accuracy of image feature extraction and reduce the error in the pose estimation process, we design a feature extraction algorithm based on skipfusion-FCNN. The traditional fully convolutional neural network (FCNN) is improved, a skip-fusion-FCNN network model is proposed, and three different paths are constructed for feature extraction. In each path, the prediction results of different depths are fused by downsampling to obtain a feature map. Merge three different feature maps to obtain feature fusion information, taking into account the structural information and detail information of the image. Experiments show that this algorithm is superior to the state-of-the-art algorithms.

Citation Download Citation

Bo Chen, Kun Yan, Rongchuan Cao, Tianqi Zhang, and Xiaoli Zhang "A new visual odometry algorithm based on multi-path deep fully convolutional neural networks", Proc. SPIE 12474, Second International Symposium on Computer Technology and Information Science (ISCTIS 2022), 124740W (8 December 2022); https://doi.org/10.1117/12.2653846

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
9 PAGES

DOWNLOAD PAPER SAVE TO MY LIBRARY

GET CITATION

RIGHTS & PERMISSIONS

Get copyright permission Get copyright permission on Copyright Marketplace

KEYWORDS

Image fusion

Visualization

Feature extraction

Convolutional neural networks

Image quality

Evolutionary algorithms

Image processing

Show All Keywords

Keywords/Phrases

Search In:

Publication Years