Recent Researches on Electronics and Earth Sciences International Conference, Turkey, 1 - 04 October 2001, pp.63-75
In this paper, Random Neural Network (RNN) has been applied to a real archaeological magnetic map belonging to ancient Hitit civilization, obtained by an international research group. RNN is a contemporary, stochastic approach for 2-D image processing for real-time models. The domain-dependent prior knowledge, such as the sizes, the shapes, the depth and the orientations of observed regions can be reflected in the parameters of RNN structure. We have separated regional and residual anomalies of Hitit civilization using RNN and extracted the historical Sarissa -Kusakli walls close to Sivas city in Turkey. We have shown that the performance of RNN is better than classical derivative based techniques.