4-D pattern structure features by three stages clustering algorithm for image analysis and classification

Abstract

An approach for decomposition of visual image by clustering, pattern analysis and classification by structure features is considered. Hierarchical clusters such as rectangles, closed regions and integrated areas are objects of investigation. By hierarchically constructed fragments, the 4-D pattern structure features are formulated. To reduce the clustering algorithm complexity, the scanning area approach is proposed. The results of pattern analysis and classification by structure features for some images are presented.

DOI: 10.1007/s10044-013-0326-x

Extracted Key Phrases

16 Figures and Tables

Cite this paper

@article{Melnyk20134DPS, title={4-D pattern structure features by three stages clustering algorithm for image analysis and classification}, author={Roman Melnyk and Ruslan Tushnytskyy}, journal={Pattern Analysis and Applications}, year={2013}, volume={16}, pages={201-211} }