Evaluation of Infrared Target Discrimination Algorithms.
Final rept. May 82-Feb 83,
INSTITUTE FOR DEFENSE ANALYSES ALEXANDRIA VA SCIENCE AND TECHNOLOGY DIV
Pagination or Media Count:
This paper is concerned with the evaluation of algorithms used by passive infrared sensors to discriminate between signals due to target sources and those due to background clutter. The discussion is essentially restricted to the case of point targets. The goal is to obtain a rough estimate of performance against minimum standards. For this purpose the analysis assumes a simple mathematical model for the background clutter distribution namely, that it is multivariate Gaussian over the spatial and spectral data channels provided by the sensor. The paper also discusses experimental evidence for and against such a model, as well as certain more explicit statistical models that have been proposed for the spatial distribution of clutter. Other topics discussed are CFAR optimum processing, linear filters, the effect of using ratios of spectral components for processing in multi-color systems rather than the components, themselves, and background normalization. Also discussed is the relationship between the effectiveness of tracking algorithms and the preliminary screening of targets by CFAR detection algorithms.
- Theoretical Mathematics
- Infrared Detection and Detectors