Development of Stereo Visual Odometry Based on Photogrammetric Feature Optimization

Sung-Joo Yoon,Taejung Kim

doi:10.3390/rs11010067

Sung-Joo Yoon, Taejung Kim

Open Access

https://doi.org/10.3390/rs11010067

Copy DOI

Journal: Remote Sensing	Publication Date: Jan 1, 2019
Citations: 6	License type: CC BY 4.0

Affiliation: Inha University

Abstract

One of the important image processing technologies is visual odometry (VO) technology. VO estimates platform motion through a sequence of images. VO is of interest in the virtual reality (VR) industry as well as the automobile industry because the construction cost is low. In this study, we developed stereo visual odometry (SVO) based on photogrammetric geometric interpretation. The proposed method performed feature optimization and pose estimation through photogrammetric bundle adjustment. After corresponding the point extraction step, the feature optimization was carried out with photogrammetry-based and vision-based optimization. Then, absolute orientation was performed for pose estimation through bundle adjustment. We used ten sequences provided by the Karlsruhe institute of technology and Toyota technological institute (KITTI) community. Through a two-step optimization process, we confirmed that the outliers, which were not removed by conventional outlier filters, were removed. We also were able to confirm the applicability of photogrammetric techniques to stereo visual odometry technology.

Highlights

Estimation of a platform’s pose using a sensor is a technology that has attracted attention in various fields, such as robotics and the automobile industry
VO is divided into monocular visual odometry (MVO) and stereo visual odometry (SVO)
We showed that the proposed photogrammetric processing could enable successful outlier removal and that real-time processing was feasible even with photogrammetric iterative estimations

Summary

Introduction

Estimation of a platform’s pose using a sensor is a technology that has attracted attention in various fields, such as robotics and the automobile industry. Typical sensors include the global positioning system (GPS), light detection and ranging (LiDAR), and the camera. The GPS is the most popular method, and sub-meter accuracy is possible. Accurate GPS equipment is very expensive, and accuracy is greatly reduced in some environments where satellite signals are blocked, such as downtown or in tunnels [1]. The method using LiDAR is very accurate and stable. Since it requires expensive equipment, its application is limited. The method using a camera has a great advantage that the construction cost is relatively low. This technique is called visual odometry (VO). VO is divided into monocular visual odometry (MVO) and stereo visual odometry (SVO)

Methods

Results

Discussion

Conclusion