3D Reconstruction of Small Solar System Bodies using Rendered and Compressed Images

Loading...
Thumbnail Image
Journal Title
Journal ISSN
Volume Title
Sähkötekniikan korkeakoulu | Master's thesis
Date
2020-03-16
Department
Major/Subject
Space Robotics and Automation
Mcode
ELEC3047
Degree programme
Erasmus Mundus Space Master
Language
en
Pages
82+3
Series
Abstract
Synthetic image generation and reconstruction of Small Solar System Bodies and the influence of compression is becoming an important study topic because of the advent of small spacecraft in deep space missions. Most of these missions are fly-by scenarios, for example in the Comet Interceptor mission. Due to limited data budgets of small satellite missions, maximising scientific return requires investigating effects of lossy compression. A preliminary simulation pipeline had been developed that uses physics-based rendering in combination with procedural terrain generation to overcome limitations of currently used methods for image rendering like the Hapke model. The rendered Small Solar System Body images are combined with a star background and photometrically calibrated to represent realistic imagery. Subsequently, a Structure-from-Motion pipeline reconstructs three-dimensional models from the rendered images. In this work, the preliminary simulation pipeline was developed further into the Space Imaging Simulator for Proximity Operations software package and a compression package was added. The compression package was used to investigate effects of lossy compression on reconstructed models and the possible amount of data reduction of lossy compression to lossless compression. Several scenarios with varying fly-by distances ranging from 50 km to 400 km and body sizes of 1 km and 10 km were simulated and compressed with lossless and several quality levels of lossy compression using PNG and JPEG 2000 respectively. It was found that low compression ratios introduce artefacts resembling random noise while high compression ratios remove surface features. The random noise artefacts introduced by low compression ratios frequently increased the number of vertices and faces of the reconstructed three-dimensional model.
Description
Supervisor
Praks, Jaan
Thesis advisor
Slavinskis, Andris
Keywords
small solar system bodies, compression, image rendering, deep space exploration, simulation, computer vision
Other note
Citation