<title>Background suppression and tracking with a staring mosaic sensor</title>

Abstract

This paper presents theoretical analysis for a staring mosaic infrared sensor with representative examples of data processing from a computer simulation. The analysis treats: (1) generation of synthetic two-dimensional scenes with specified cloud geometry and desired statistical characteristics, (2) the processing of frames of data from two-dimensional scenes to represent temporal, spatial, and multispectral filtering, and (3) the thresholding and examination of the processed scenes to implement track association. The temporal filtering includes multiple differencing, statistically optimal non-recursive filtering, and recursive filtering. Methods are presented for reducing the computation load when calculating the optimal coefficients in spatial and multispectral filtering. The track association uses thresholding and examination to eliminate stationary objects with track assembly similar to the "streak algorithm." For visual display, the two-dimensional scenes and the processed frames are output with a forty-eight level gray scale.