Application of Artificial Intelligence in Classification of Maritime Targets

Fast classification of a target according to RCS signals is important for many applications. In this paper, we describe the use of the Neural Networks for object classification using collected RCS real data from radar system. Our collected RCS polar plots for 3 ship classes are applied to NNs. This paper proposes three models of three layered feed-forward Neural Network and back-propagation training algorithm. In the first one, we feed 75 inputs to the NN which are frequency, polarization, 72 RCS values and the mean of these values. This method gives 100% overall correct classification. In the second one, we feed four inputs to the NN which are frequency, polarization, aspect angle and its corresponding RCS value. This method gives 76% overall correct classification but it is more efficient in actual scenario because there is no guarantee to view a full revolution of the object. In the third model we introduce six inputs to the NN which are the same inputs of the second model besides adjacent aspect angle and its corresponding RCS value, consequently we obtained 84% overall correct classification. We can use this model when the whole RCS of the object is unavailable.