Vision-based Spacecraft Trajectory Estimation Using Structure from Motion

3D reconstruction of asteroid Bennu's surface and spacecraft trajectory estimation using SfM techniques.

Overview

This project focuses on the 3D reconstruction of asteroid Bennu’s surface using Structure from Motion (SfM) techniques, as well as the estimation of a spacecraft’s trajectory during the OSIRIS-REx mission. By leveraging feature detection (using SIFT) and robust outlier mitigation (using RANSAC), our approach reconstructs the scene in three dimensions while simultaneously determining the camera/spacecraft poses. The following results showcase our custom incremental SfM implementation alongside OpenMVG’s incremental and global SfM approaches.


Custom Incremental SfM Implementation

The figures below illustrate our own implementation of incremental SfM. The left image displays the feature matches between two successive images, while the right image visualizes the 3D reconstruction generated by our algorithm.

Custom Incremental SfM: Feature matches (left) and 3D reconstruction (right).

OpenMVG Implementations

In addition to our custom approach, we have implemented both incremental and global SfM using the OpenMVG library. The figures below present the results obtained from these implementations.

OpenMVG Implementations: Incremental SfM (left) and Global SfM (right).