BDBComp
Parceria:
SBC
Texture Classification Using the Lempel-Ziv-Welch Algorithm

Leonardo Vidal BatistaMoab Mariz Meira

This paper presents a new, simple and efficient texture classification method using Lempel-Ziv-Welch (LZW) compression algorithm. In the learning stage, LZW algorithm constructs dictionaries for the horizontal and vertical structure of each class. In the classification stage, texture samples to be classified are encoded by LZW in static mode, using the dictionaries constructed in the learning stage, in vertical and horizontal scanning order. A sample is assigned to the class whose dictionaries minimize the average horizontal and vertical coding rate. The classifier was evaluated for various sample sizes and training set sizes, using 30 Brodatz textures. The proposed method correctly classified 100% of 3000 Brodatz texture samples, and direct comparisons indicated the superiority of the method over several high performance classifiers.

http://springerlink.metapress.com/content/g40ln7499phh9y1k/?p=a7e6524c664448e48a1d2e5c77c28c11&p

Caso o link acima esteja inválido, faça uma busca pelo texto completo na Web: Buscar na Web

Biblioteca Digital Brasileira de Computação - Contato: bdbcomp@lbd.dcc.ufmg.br
     Mantida por:
LBD