Date of Award

2023

Degree Type

Thesis

Degree Name

Master of Science in Ocean Engineering

Department

Ocean Engineering

First Advisor

Chris Roman

Abstract

Navigating the underwater environment is a challenging task for modern visual odometry algorithms due to poor visibility and a lack of distinct points for feature-based match making. This research increases the navigation capability of undersea vehicles via the adoption of a more robust, featureless image registration algorithm for visual odometry. Featureless image registration utilizes the frequency domain in order to match images based on their cross power spectrum, and therefore does not require distinct feature points. As such, this method displayed reliable matching in environments with low texture, improper lighting, high contrast, and high texture, and has a consistent computational cost capable of real-time use on low cost systems such as the raspberry pi 4B+ microcomputer. In addition, low cost camera technologies were also utilized to demonstrate their ability to reduce barriers to entry for autonomous ocean research. Results demonstrated satisfactory velocity estimation and reliable altitude measurements utilizing calibrated stereo cameras. The proposed method builds off a number of works encouraging the use of featureless based image registration algorithms in the undersea environment as opposed to more traditional algorithms, and opens up avenues for future work in enhanced localization capabilities for undersea vehicles.

Share

COinS
 
 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.