Comparison Between a Non-Linear and a Linear Gaussian Statistical Detector for Detecting Dim Satellites.

摘要

This paper describes and analyzes a Gaussian statistical satellite detection algorithm. This detection algorithm exploits the correlation of pixels that arises from the long-term exposure point spread function (PSF) due to diffraction and atmospheric effects. This correlating detection algorithm is compared to a simple, linear threshold detection algorithm, which treats the probability density function (PDF) of the photons as Gaussian and the PSF as Dirac. The effects of changing the window size and grid size is explored and analyzed. Using a dataset collected by the Space Surveillance Telescope (SST) that shows many samples of a geostationary satellite gradually going into eclipse during the vernal equinox, the probabilities of detection for the algorithms are compared as the satellite becomes very dim and nearly disappears to determine which detection algorithm performs better. A better performing detection algorithm will allow detection of satellites, space debris, and other dim objects that have not been visible before.