Busca avançada
Ano de início
Entree
(Referência obtida automaticamente do Web of Science, por meio da informação sobre o financiamento pela FAPESP e o número do processo correspondente, incluída na publicação pelos autores.)

Improving texture extraction and classification using smoothed morphological operators

Texto completo
Autor(es):
Neiva, Mariane Barros [1, 2] ; Vacavant, Antoine [3] ; Bruno, Odemir Martinez [1, 2]
Número total de Autores: 3
Afiliação do(s) autor(es):
[1] Univ Sao Paulo, Sao Carlos Inst Phys, Sci Comp Grp, POB 369, BR-13560970 Sao Carlos, SP - Brazil
[2] Univ Sao Paulo, Inst Math & Comp Sci, Trabalhador Sao Carlense Ave, 400-Downtown, BR-13566590 Sao Carlos, SP - Brazil
[3] Univ Clermont Auvergne, CNRS, Inst Pascal, UCA, SIGMA, UMR6602, F-63171 Aubiere - France
Número total de Afiliações: 3
Tipo de documento: Artigo Científico
Fonte: DIGITAL SIGNAL PROCESSING; v. 83, p. 24-34, DEC 2018.
Citações Web of Science: 1
Resumo

To improve texture recognition, this paper proposes the application of a morphological transformation in the original dataset. The smoothed shock filtering method uses smoothed dilations and erosions operations to produce enhanced images respecting its edges. As the preprocessing method is applied iteratively, it produces a scale space representation, which allows us to analyze combinations of them to check the ones that give the best results for texture recognition. To describe transformed and original images, six different texture analysis methods will be used. Results of the proposed method are compared with the original approach (without any preprocessing method) and the use of two classic diffusion algorithms. Achievements with our smoothed shock filter show a classification rate of 99.54% for Vistex classification using CLBP and KNN (k = 1) as the classifier. (C) 2018 Published by Elsevier Inc. (AU)

Processo FAPESP: 14/08026-1 - Visão artificial e reconhecimento de padrões aplicados em plasticidade vegetal
Beneficiário:Odemir Martinez Bruno
Linha de fomento: Auxílio à Pesquisa - Regular