2D Image pre-processing — Part I: Fuzzy 3D interpolation

This paper analyses the possibilities to make a spatial image pre-processing using a fuzzy 3D interpolation algorithm. The proposed algorithm computes the crisp estimation using fuzzy inference rules for information contained in neighbouring pixels. If the membership functions are chosen of linear type, then the crisp estimation for the basic 3D interpolation fuzzy algorithm of zero level, which uses only information regarding the nearest four pixels, is the same as estimation relation used by the bilinear 3D algorithm. Using a knowledge-based model, the fuzzy 3-D interpolation that uses more information from the closest pixels can give a natural solution for the surface of estimation. The fuzzy 3-D interpolation model of first level that uses 12 pixels and linear membership functions gives an estimation surface of third order.