• Refine Query
  • Source
  • Publication year
  • to
  • Language
  • 1
  • Tagged with
  • 2
  • 2
  • 2
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • About
  • The Global ETD Search service is a free service for researchers to find electronic theses and dissertations. This service is provided by the Networked Digital Library of Theses and Dissertations.
    Our metadata is collected from universities around the world. If you manage a university/consortium/country archive and want to be added, details can be found on the NDLTD website.
1

Image Processing Pipeline and a Global Search for Local Maxima Method of Object Detection

Nathan Kampe Houtz (19273654) 01 August 2024 (has links)
<p dir="ltr">Optical observations provide a cost-effective means of tracking satellites and space debris with high angular accuracy, and many amateur, academic, and professional observers use them extensively. Optical images can be accumulated quickly, and automation is important to rapidly produce accurate measurements of objects found in them. Effects like atmospheric refraction, atmospheric scattering of incoming light, aberration due to the motion of the observer, image distortions from the optics of the telescope, scintillation due to atmospheric turbulence, limited resolution, and various sources of noise create challenges for observers. An image processing pipeline has been developed from scratch for the purpose of automating the collection of data with the Purdue Optical Ground Station telescope. Effects that are deterministic are mathematically modeled and corrected, and all steps of the pipeline are described. A novel method is presented for detecting and optimally estimating the centroids of faint, streaked objects in astronomical images with several-second-long exposure times. The ability to accurately determine the pointing direction of a telescope from the stars in the image is demonstrated with a series of images of a GPS satellite. The resulting orbit is compared with the broadcast ephemeris, with an average positional error of 22.1 meters over the observation period.</p>
2

Feasibility study of initial orbit determination with open astronomical data / Studie av initial banbestämning med öppen astronomisk data

Mattsson, Linn January 2022 (has links)
In this report I present a feasibility study of using open astronomical data to make Initial Orbit Determination (IOD) for Resident Space Objects (RSO) appearing as streaks in telescope images. The purpose is to contribute to Space Surveillance and Tracking (SST) for maintaining Space Situation Awareness (SSA). Data from different wide-field survey telescopes were considered but due to availability constraints only mask images from Zwicky Transient Facility (ZTF) survey were chosen for the analysis. An algorithm was developed to detect streaks in the mask images and match them to RSO known to be within the Field of View (FoV) at the observation time. Further, the IOD was made with angles-only Laplace’s method and the state vectors calculated for the streaks from the IOD were compared to those from the TLE for the matching RSO. The algorithm was tested with 6 different image fields acquired between the 14th to the 16th December 2019, of which 4 are characterised as non-crowded and 2 as crowded. The streak finding algorithm has a better precision and sensitivity for the non-crowded field, with an F1-score of 0.65, but is worse for the crowded fields with an F1-score of 0.035. In the non-crowded fields 95% of all streak and object matches are true matches to unique RSO, while for the crowded field only 10% are true matches. It was found that the 1''/pixel resolution in the images is too low for doing an IOD with Laplace’s method, despite how well the streak finding algorithm performs. However, with some improvements, the method is suitable as a cost effective way to verify known RSO in catalogues. / I den här rapporten presenterar jag en studie om att använda öppen astronomiska data för att göra initial banbestämning för artificiella rymdobjekt avbildade som streck i teleskopbilder. Syftet är att tillhandahålla information för att upprätthålla en god rymdlägesbild. Data från olika kartläggnings teleskop övervägdes men på grund av begränsningar i tillgänglighet valdes endast mask-bilderna från Zwicky Transient Facility för analysen. En algoritm utvecklades för att upptäcka streck i mask-bilderna och matcha dem med kända objekt i bildens synfält vid observationstillfället. Vidare gjordes den initiala banbestämningen med Laplaces metod, som använder vinkelkoordinaterna för streckens position vid observationen. Tillståndsvektorerna för strecken och de matchade objekten jämfördes, de beräknades från den initiala banbestämningen respektive objektets TLE. Algoritmen testades med 6 olika bildfält från observationsdatum mellan den 14:e till den 16:e december 2019, av dessa karakteriseras 4 som glesa och 2 som fyllda. Algoritmen för streck detektering har bättre precision och känslighet för de glesa fälten, med ett F1-värde på 0.65, men sämre för de fulla fälten med ett F1-värde på 0.035. I de glesa fälten är 95% av alla streck- och objektmatchningar korrekta matchningar med unika objekt, medan för det fulla fälten är endast 10% korrekta matchningar. Det visar sig att upplösningen på 1''/pixel i bilderna är för låg för att göra en initial banbestämning med Laplaces metod, oavsett hur bra algoritmen för streck detektering presterar. Genom att göra vissa förbättringar i algoritmen är metoden lämplig för att, på ett kostnadseffektivt sätt, verifiera kända objekt i kataloger.

Page generated in 0.0869 seconds