Grid star identification improvement using optimization approaches
| Autor: | Hamid Abrishami Moghaddam, Mahdi Aghaei |
|---|---|
| Rok vydání: | 2016 |
| Předmět: |
Computer science
media_common.quotation_subject Population ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION Aerospace Engineering 02 engineering and technology Star position 01 natural sciences Standard deviation Apparent magnitude 0203 mechanical engineering Robustness (computer science) 0103 physical sciences Computer vision Electrical and Electronic Engineering education 010303 astronomy & astrophysics Radiometric calibration media_common 020301 aerospace & aeronautics education.field_of_study Pixel business.industry Grid Sky Radiometry Radiometric dating Artificial intelligence business Algorithm |
| Zdroj: | IEEE Transactions on Aerospace and Electronic Systems. 52:2080-2090 |
| ISSN: | 0018-9251 |
| DOI: | 10.1109/taes.2016.150053 |
| Popis: | The grid method is an all-sky star identification approach which tolerates high values of position and stellar magnitude noise compared to its state-of-the-art counterparts. Nevertheless, its performance is affected by sensor’s radiometric bias and large values of position noise standard deviation (>1 pixel). In this paper, some optimization approaches are exploited to make the grid algorithm more robust against the above hazards. First, a statistical analysis of stars’ population with respect to the stellar magnitudes is performed to define radiometric clusters which make the grid algorithm less sensitive to radiometric calibration bias. Second, by optimizing an objective function defined by the product of robustness factor and grid resolution, the optimal grid cell size is achieved to make the algorithm more robust against noisy sky field. Finally, using Bayesian decision theory, a method is developed for computing optimal static threshold for miss/correct matches classification. The proposed algorithm has been tested on a camera (with 14.6◦ ×14.6◦ field of view and 512×512 pixels resolution) against harsh conditions of both star position and stellar magnitude uncertainty. Sensor identification probability of 99.8% against 2 pixels position noise standard deviation has been obtained. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|