Date of Award

Fall 2025

Document Type

Open Access Thesis

Department

Mechanical Engineering

First Advisor

Yi Wang

Abstract

This work presents a camera system for estimating surface currents. A combined camera and IMU system provide high resolution images and high precision orientation information, respectively. This eliminates the need for leveling and enables forward looking operation. The hardware stack includes a Lucid Vision Triton TRT162S monochrome camera, an SBG Ellipse D with dual antenna GNSS for precise pose, and an Nvidia Jetson Orin for acquisition and processing. The Ellipse D provides a hardware trigger at 10 Hz to time stamp each frame and to synchronize images with inertial measurements.

A camera calibration step estimates the intrinsic camera parameters and distortion coefficients. Images are undistorted and then orthorectified to the water surface using measured orientation and a projective transformation. A distance-based cropping method limits the field of view to a valid world range and improves numerical stability. The orthorectification is accelerated by mapping four cropped corners rather than projecting every pixel, which enables about seven frames per second on a desktop CPU and provides headroom for GPU use.

Surface currents are recovered from the orthorectified sequence by forming frequency and wavenumber spectra over a fixed region of interest in world coordinates. A two-dimensional and a three-dimensional curve fitting approach align theoretical dispersion relations with the measured spectra to estimate the current vector. Orthorectification accuracy is within about ten percent across a range of approximately 15 meters. Field trials in Charleston Harbor produce near and mid-range current vectors at approximately 50 meters and 85 meters, respectively. The results demonstrate an easily deployable system for surface current measurements. Future work will test this system on moving platforms (small vessels) and incorporate current measurements analysis to the Nvidia Jetson for close to real time measurements.

Rights

©2025, Colby James Weeks

Download

Share

COinS