A fuzzy topological framework for classifying image databases

✍ Scribed by Homa Fashandi; James F. Peters

• Publisher: John Wiley and Sons
• Year: 2011
• Tongue: English
• Weight: 443 KB
• Volume: 26
• Category: Article
• ISSN: 0884-8173
• DOI: 10.1002/int.20479

✦ Synopsis

The problem considered in this paper is how to classify image databases in terms of semantically coherent image categories. An image category (image concept) is represented by a set of images with visual and semantic similarities. We propose a topological framework to model each image concept and also classify images. Classification part utilizes a new form of fuzzy interior. To cope with uncertainties associated with feature extraction part and to achieve high classification acuity, the proposed fuzzy interior method takes into account the inclusion degree. This is accomplished by considering a variation of fuzzy subsethood used in the definition of fuzzy interior and by viewing image feature histograms as fuzzy sets. Because each image category is modeled independently, adding new categories to the system is both efficient and easily accomplished. The main contribution of this paper is the introduction of a fuzzy topological framework to classify image databases.