ePrints@IIScePrints@IISc Home | About | Browse | Latest Additions | Advanced Search | Contact | Help

Principal components of Gabor decomposition for texture segmentation

Venkatesh, YV and Parthasarathy, M (1995) Principal components of Gabor decomposition for texture segmentation. In: Proceedings of Second Asian Conference on Computer Vision: ACCV '95, Vol.3, 5-8 December 1995, Singapore, pp. 762-766.

Full text not available from this repository. (Request a copy)
Official URL: http://en.scientificcommons.org/36809367


The paper deals with the problem of texture boundary detection or, equivalently, texture segmentation of images. A novel Gabor function-based technique has been proposed to solve the problem for a wide variety of textural images. The Gabor framework (involving bi-orthogonal functions) enables the decomposition of the image into multiple, independent layers containing spatial and spectral information with minimal uncertainty. This is believed to result in a better localization of textural boundaries than is possible using the methods (of the literature) not based on Gabor decomposition. An attempt is made to optimize the number of layers needed for image reconstruction by invoking the method of principal components. Using this set of `optimal' layers, a clustering algorithm has been used to group elements of the image having similar gradient values in this set of optimal layers. For lack of space, all the details are omitted, and only two examples are given in the final form in order to demonstrate the superiority of the proposed technique.

Item Type: Conference Paper
Publisher: Nanyang Technological University
Additional Information: Copyright of this article belongs to Nanyang Technological University.
Keywords: edge detection;image reconstruction;image segmentation;image texture
Department/Centre: Division of Electrical Sciences > Electrical Engineering
Date Deposited: 17 Sep 2007
Last Modified: 11 Jan 2012 05:45
URI: http://eprints.iisc.ac.in/id/eprint/10906

Actions (login required)

View Item View Item