Fast selection of the sea clutter preferential distribution with neural networks

Studies performed on sea clutter readings often include fitting the data searching for the preferential amplitude distribution. In this process, the estimation through the method of moments and the Kolmogorov-Smirnov test are usually used with positive results. However, the procedure cannot be directly applied in the fast selection of the distribution in operational schemes because it consumes a high amount of computational resources. The authors found a new way of estimating the sea clutter preferential distribution by using a neural network that takes histograms of the readings as an input, achieving faster and more precise results than the traditional alternative. The effectiveness of the proposal was verified with computer generated data for the Weibull, Log-Normal and K distributions. Besides, analyses were executed including real radar samples taken with the IPIX radar.