Local Fuzzy Pattern: A New Way for Micro-pattern Analysis

The aim of this paper is to introduce a new methodology for micro-pattern analysis in digital images. The gray-level pixels’ structure in an image neighborhood describes a spatial specific context. Edge, line, spot, blob, corner or texture can be described by this structure. The gray-level values of the image pixel are interpreted as a fuzzy set, and each pixel gray-level as a fuzzy number. A membership function can be defined to describe the membership degree of the central pixel to the others in an image neighborhood. We have called this method the Local Fuzzy Pattern (LFP). If a sigmoid membership function is used, the proposed methodology describes the texture very well, and if a symmetrical triangular membership function is applied, the LFP is better for edge's detection. The results were compared to the Local Binary Pattern (LBP), for texture classification getting the better hit-rate. Our proposed formulation for the LFP is a generalization of previously published techniques, such as Texture Unit, LBP, FUNED, and Census Transform.