Signal Sequence Detection Given Common Background Image Count Sets

The optimum processing is found to detect a signal sequence in a set of M "images" with a common background; the "images" are the final counts, after a fixed reception time, for each of J spatially separate detectors and for each of M separate receptions. The spatially varying background is assumed either unknown or normal with known mean and covariance; results are generally confined to the device noise limited case. It is shown that, unlike the classical case, optimum signal selection, along with optimum processing, can not entirely eliminate the detectability loss due to the background.