Texture analysis based on maximum contrast walker

Author(s)

Backes, Andre Ricardo ; Martinez, Alexandre Souto ; Bruno, Odemir Martinez

• Document type: Journal article
• Source: PATTERN RECOGNITION LETTERS; v. 31, n. 12, p. 7-pg., 2010-09-01.

Abstract

Recently, the deterministic tourist walk has emerged as a novel approach for texture analysis. This method employs a traveler visiting image pixels using a deterministic walk rule. Resulting trajectories provide clues about pixel interaction in the image that can be used for image classification and identification tasks. This paper proposes a new walk rule for the tourist which is based on contrast direction of a neighborhood. The yielded results using this approach are comparable with those from traditional texture analysis methods in the classification of a set of Brodatz textures and their rotated versions, thus confirming the potential of the method as a feasible texture analysis methodology. (C) 2010 Elsevier B.V. All rights reserved. (AU)