This paper proposes a new approach to railway vehicles localization based on image processing method consists in the recognition of the railway car coupler on the graphical images obtained by cameras of versatile automatic recognition system for railway cars’ numbers (versatile ARS-RCN system). A model based on the recognition of the coupling of the real-valued negative selection algorithm with variable-sized detectors was developed. Computational experiments that showed the effectiveness of developed model in comparison with the classical support vector machine on real data collected on one the versatile ARS-RCN system object were performed. The effectiveness of the proposed approach in the duplicated mode (using two cameras) and in combination with the basic method of localization of mobile units on the basis of inductive sensors passing wheels was showed.
Keywords: artificial immune system, soft computing, localization of railway vehicles , identification of cars’ numbers