Estimation of Shape and Relative Motion for Partially Resolved Objects in Optically Acquired Imagery

Abstract

This work aimed to advance 2D image and 3D mesh processing techniques for partially resolved space objects. Over the course of this project, we made advancements in the following areas: construction of a 3D scale space and finding salient features (keypoints) in this scale space, graph-based methods for matching keypoints of a newly observed object with a catalog of known objects, 3D shape modeling from sequences of silhouette images, recognition of a partially resolved object in a 2D image with a convolutional neural network, analytic corrections for bearing bias of partially resolved objects.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
May 28, 2019
Accession Number
AD1096751

Entities

People

  • Andrew P Rhodes
  • Ashish Jagat
  • Christopher A. Ertl
  • Jacob Puritz
  • Jason Gross
  • John A Christian

Organizations

  • West Virginia University

Tags

Communities of Interest

  • Air Platforms
  • Autonomy
  • Space

DTIC Thesaurus Topics

  • Air Force Research Laboratories
  • Algorithms
  • Artificial Intelligence Software
  • Artificial Satellites
  • Computer Vision
  • Convolutional Neural Networks
  • Curvature
  • Department Of Defense
  • Identification
  • Image Processing
  • Machine Learning
  • Navigation
  • Neural Networks
  • Object Recognition
  • Recognition
  • Space Objects
  • Space Surveillance

Readers

  • Computer Vision.
  • Neural Network Machine Learning.

Technology Areas

  • AI & ML
  • AI & ML - Machine Learning Algorithms
  • Space
  • Space - Space Objects