Local Fuzzy Color Patterns For Image Retrieval

摘要

This paper proposes two extensions of the classical Local Binary Pattern (LBP) method, that has been previously used for scalar (gray-level) texture and image recognition tasks. The first extension embeds fuzzy logic into the definition of the LBP. The second extension allows the use of the new descriptor, named Local Fuzzy Color Pattern (LFCP) for color (vector-valued) images. Our approach is based on the introduction of a fuzzy color similarity function, such that the LFCP descriptor embeds information on the spatial color distribution within a first-order statistical distribution. Experiments show that the proposed descriptors can be successfully used for color image recognition and retrieval.