Space Object Detection Algorithm Leveraging Absolute Photo-Detector Calibration

This paper introduces a new space object detection algorithm that is designed to process image data taken from astronomical telescopes for the purposes of finding sources of optical radiation in space. Specifically, the algorithm is designed to find unresolvable space objects or objects that possess an angular size that is too small to appear as anything, but a point source as viewed through the telescope conducting the search. The proposed approach involves calibrating the image data into units of photoelectrons and then executing an estimation algorithm to compute the strength of the hypothetical sources in the image. A Likelihood Ratio Test (LRT) is then implemented to make a determination if the hypothetical sources are classified as space objects or not. The proposed algorithm is demonstrated to achieve a higher probability of detecting unresolvable objects than the matched filter, which is still the state-of-the-art approach for finding optical sources in astronomical images. The new approach involves a pre-processing step where the amplitude of the optical source in a given test location is estimated under the hypothesis that at optical source exists at that location. The median filter is used to estimate the background level in the vicinity of the test location. Once these parameters are estimated, A likelihood ratio test is used to determine whether an object is present at the test location. The new algorithm is tested against the matched filter detector using two sets of measured short exposure data of the star Polaris and two stars in its vicinity taken with a small telescope. Receiver Operating Characteristic (ROC) curves are produced for the two detection schemes showing that the new algorithm out-performs the old one with a difference of 10 percent in the probability of detection, which is demonstrated to be statistically significant in these experiments with confidence as high as 90%.