Reinterpreting the CLEAN algorithm as an optimum detector

This paper gives a short review of the CLEAN algorithm and shows how it can be implemented as a deconvolver in place of a correlator. The paper then derives an optimum detector (with appropriate assumptions) that provides detection of closely spaced Rayleigh distributed objects in the presence of additive white Gaussian noise (AWGN). These two processors are shown to both be deconvolvers for infinite signal to noise ratio. The range-time cut of the ambiguity diagram indicates that both algorithms have outstanding range-time sidelobes (under the condition of infinite signal to noise ratio). The optimum detector is not a uniformly most powerful test because it incorporates target amplitude into the detector. However, in many practical situations estimates of this parameter may be available.