Image Processing Technique for Automatic Detection of Satellite Streaks
Abstract
This report presents an image processing technique to detect satellite streaks. This method is particularly useful in the context of surveillance of space where the positions of active satellites and other orbital debris must be monitored. In such cases, the orbital parameters are known, but after a certain time this knowledge deteriorates because of the recent orbit perturbations. Consequently, the satellites need to be re-observed and the orbital parameters must be updated before these satellites are declared lost. Hence, even before its reacquisition, the satellite's motion is known but its position is estimated with a certain margin of error. This knowledge allows for automatic pointing of the telescope, acquiring of images with the satellite streak within the field of view, developing a matched filter (with the motion knowledge) for detecting this streak within the image and declaring of the satellite position. This study begins with a detailed analysis of a typical image, which includes several sensor artefacts (such as dead pixels, background gradient, noise) and signal degradation (bleeding, blooming, saturation, etc.). This study explains how sensor artefacts are corrected, image background is removed and noise is partially erased. Then, it describes a technique to detect and erase stars. This reduces the number of objects in the image that could generate false alarms. Finally, this document describes how to design and apply a matched filter used for extracting the satellite streak in images. Examples of processed data are illustrated for each of the processing steps.
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 01, 2007
- Accession Number
- ADA596420
Entities
People
- Martin P. Levesque
- Sylvie Buteau
Organizations
- DRDC Valcartier