Performance analysis of closely spaced objects resolution algorithms

Abstract

Two leading algorithms for resolving closely spaced objects (CSOi are discussed and their performance in terms of the probability of CSO resolution i s compared through computer simulation. Algorithms considered are the maximum likelihood (ML) estimation with Akaike criterion and deconvolution filter (DF) based upon autoregressive modeling of the system response function. Computer simulation shows that the ML estimation outperforms the DF method with respect to the probability of resolution on two CSO. However, as for the number of computations and the comelexity of operations. the DF is simpler and more efficient to run in real-time, since the number of operations of the ML method depends on the number of CSO to be resolved, while the DF method is not affected. Simulation results are presented for the two dimensional (2-0) case with Gaussian noise.