Prototype of automatic satellite streak detection, identification and initial orbit determination pipeline from optical observation

Abstract

We present a working prototype of automatic satellite streak detection, identification and initial orbit determination pipeline from optical telescope images based on python. The pipeline process calibrated and plate-solved astronomical images in fits format. We utilized Hough transform-based feature detection to detect the light streaks and extract the equatorial position based on the image header information. The streaks were then identified by comparing their center position with the calculated SGP4 based satellite position from available TLE data. The corresponding initial orbit is calculated by utilizing the Laplacian method by interpolating the streak position and time. We test the pipeline by using 85 blind-survey 10 cm telescope images around the celestial equator. The pipeline correctly detects 123 lines with 0.85 sensitivity, automatically identify 92 detected satellites unambiguously and determine the initial orbit of 118 streaks.