Detection in dense signal environments via the Gabor transform

The matched filter is the classical solution for detection of an isolated, known signal in Gaussian noise. In dense signal environments consisting of numerous overlapped pulses, the matched filter is not necessarily the optimum detector. In this paper we explore Gabor-based detectors for dense environments, in order to capitalize on the potentially favorable trade-off of detector signal-to-noise ratio (SNR/sub DET/) vs. resolution. Theory and experiments are presented that compare Gabor and matched filter detectors under various environments and support the conclusion that the Gabor detector is superior, as measured by the receiver operating characteristic, under suitably dense environments.