Title: A new approach for texture classification in CBIR

Authors: Shengju Sang, Mingxia Liu, Jing Liu, Qi An

Addresses: School of Mechanical and Power Engineering, East China University of Science and Technology, Shanghai 200237, China; and Department of Information Science and Technology, Taishan College, Taian 271021, China. ' Department of Information Science and Technology, Taishan College, Taian 271021, China. ' Department of Information Science and Technology, Taishan College, Taian 271021, China. ' School of Mechanical and Power Engineering, East China University of Science and Technology, Shanghai 200237, China

Abstract: In the field of Content-Based Image Retrieval (CBIR), the semantic understanding of textures has long been a difficult problem, especially the texture classification. This paper proposes a new approach for texture classification, which adopts ten words describing textures in natural language. Texture features of an image are extracted by Discrete Wavelet Transform (DWT), and then classified through both Back Propagating Neural Network (BPNN) and Support Vector Machine (SVM) classifiers. Experimental results show that this approach of texture classification for natural texture is feasible.

Keywords: CBIR; content-based image retrieval; texture classification; DWT; discrete wavelet transforms; BPNN; back propagation neural networks; SVM; support vector machines; WPT; wavelet package transforms; natural language; texture features; image processing.

DOI: 10.1504/IJCAT.2010.034137

International Journal of Computer Applications in Technology, 2010 Vol.38 No.1/2/3, pp.34 - 39

Received: 01 Aug 2008
Accepted: 21 Nov 2008
Published online: 16 Jul 2010 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article