Publication Details
Texture Analysis via Data Mining
Association rules, Apriori algorithm, texture analysis, texture primitive, feature vector, Fisher criterion, inter-cluster separability.
This paper deals with an idea of the use of data mining approach in texture analysis. A new method based on association rules is proposed. This approach extracts texture primitives from an image texture and describes mutual relationships between these primitives. Within this method, a technique for feature vector construction without a priori knowledge of textures different from the analyzed one is presented. The Fisher criterion was used to measure an ability of the proposed method for successful discrimination of pairs of textures. This work provides also comparison of the proposed technique with a wide-used wavelet texture features.
@inproceedings{BUT10926,
author="Martin {Heckel}",
title="Texture Analysis via Data Mining",
booktitle="Proceedings of the 2nd Australasian Data Mining Workshop",
year="2003",
pages="95--104",
publisher="University of Technology Sydney",
address="Sydney",
isbn="0-9751724-1-7"
}