Retrieving Texture Images Using Coarseness Fuzzy Partitions

Abstract. In this paper, a Fuzzy Dominant Texture Descriptor is proposed for semantically describing an image. This fuzzy descriptor is defined over a set of fuzzy sets modelling the “coarseness” texture property. Concretely, fuzzy partitions on the domain of coarseness measures are proposed, where the number of linguistic labels and the parameters of the membership functions are calculated relating representative coarseness measures (our reference set) with the human perception of this texture property. Given a “texture fuzzy set”, its dominance in an image is analyzed and the dominance degree is used to obtain the image texture descriptor. Fuzzy operators over these descriptors are proposed to define conditions in image retrieval queries. The proposed framework makes database systems able to answer queries using texture-based linguistic labels in natural language.

• Type: Journal
• Year: 2010
• Where: IPMU
• Authors: Jesús Chamorro-Martínez, Pedro Manuel Martínez-Jiménez, José Manuel Soto-Hidalgo
• Keywords: Fuzzy Dominant Texture, Information Technology, IPMU 2010, Texture Descriptor, Texture Property