Automatic extraction of linguistic models for image description

Abstract

Abstract-This paper describes a methodology to extract fuzzy models that describe linguistically the low-level features of an image (such as color, texture, etc.). The methodology combines grid-based algorithms with clustering and tabular simplification methods to compress image information into a small number of fuzzy rules with high linguistic meaning. All the steps of the methodology are carried out with the help offered by the tools of Xfuzzy 3 environment, so we can define, simplify, tune and verify the fuzzy models automatically. Several examples are included to illustrate the advantages of the methodology.

Rule Simplification Method Based on Covering Indexes for Fuzzy Classifiers

“…Finally, linguistic hedges are employed to improve the linguistic interpretability of the rules. The method was applied in [14] to extract fuzzy rules describing linguistically the low-level features (such as color, texture, etc.) of images.…”

Fuzzy Models for Fingerprint Description

“…A way to achieve this objective is to take the most significant rules for each consequent prototype (or exclude the least significant ones) as proposed in [12]. The idea is to select those rules obtained from merging more atomic rules.…”

“…Recently, there is a wide research on applying fuzzy logic to images, because images contain imprecise and ambiguous information [11]. In particular, fuzzy models of images have been done considering low-level information, such as textures or grey or color values [12]. This is the approach taken herein: the directional image will provide the texture information about the image to be modeled.…”