Use of genetic algorithms for ISAR image autofocusing

One of the most critical steps of ISAR image processing is the motion compensation, also known as ISAR image focusing. For non-cooperative targets and especially when external data are not available, autofocusing techniques must be used. Among all the techniques developed for ISAR image autofocusing, the contrast based autofocusing technique has been recently proposed by the authors. One of the critical aspects of such a technique is represented by the solution of an optimisation problem. Because the image contrast is generally a multimodal function, classic optimisation methods do not achieve the best result. In this paper a new solution of the optimisation problem is given by means of genetic algorithms. Moreover, the model of the focusing point phase history is extended to a generic polynomial and the problem of defining the polynomial order is addressed and heuristically solved. The effectiveness of the algorithm improvements, due to both the use of genetic algorithms and to the signal model extension is tested by means of real data.