Improving Satellite-Streak Detection by the use of False Alarm Rejection Algorithms

PDF

Authors
  1. Lévesque, M.P.
  2. Lelièvre, M.
Corporate Authors
Defence R&D Canada - Valcartier, Valcartier QUE (CAN)
Abstract
After having developed an image processing technique to detect satellite streaks (in a previous report), this report presents the following algorithms used to extract every alarm present in an image and reject all those that don’t look like the expected streak. The method consists first in extracting the image areas indicated by the alarms generated by detection algorithms. It then calculates the extracted-object features such as the signal-to-noise ratio and the moments of inertia. Then, the calculated features are compared with threshold values determined for the expected streak and all failed tests contribute to the rejection of false alarms. The figures of merit (probability of detection) of the detection algorithms have been calculated using simulated data, which are generated with controlled parameters. The absolute detection capability (streaks detected in presence of false alarms) and the best detection thresholds (which detect the streak without false alarms) have been established with these simulations and then confirmed with the use of real data. From an uncorrected input image, the detection algorithms can detect satellite streaks and report detections along with an estimation of false alarm probability. The excellent performance of the detection algorithms allows the detection of very faint satellite streaks that even a trained observer can hardly see.

Il y a un résumé en français ici.

Report Number
DRDC-VALCARTIER-TR-2006-587 — Technical Report
Date of publication
01 Sep 2008
Number of Pages
98
DSTKIM No
CA031356
CANDIS No
530206
Format(s):
CD ROM

Permanent link

Document 1 of 1

Date modified: