Neuro-fuzzy Model for Image Processing in Electro-optical Applications

This paper describes neuro-fuzzy approach for classification of electro optical images. It is well known that optical information is not of a big value without efficient analysis by means of classification algorithm. Even through neuro-fuzzy classification is known to give one of best results, it is still challenging for applications in underwater imaging, since ocean surface and specially foam coverage of the ocean surface contain highly reflected coming signal. It is a significant obstacle for accurate classification. Moreover, the traditional classification mapping with one-pixel-to-one-class algorithms normally fail to deal with the mixed the pixels that ordinary caused by the mixture of ocean water classes like foam. This study attempts to develop a neuro-fuzzy model with parameters that don´t have crisp determination, undefined and unknown. It is offered to use fundamentally different mathematical algorithms -the neural networks and the fuzzy logic, which are the approximators of complex (non-linear) functional dependencies for processing of optical information.