A complex network approach for dynamic texture recognition

Autor(es)

Goncalyes, Wesley Nunes ^[1] ; Machado, Bruno Brandoli ^{[1, 2]} ; Bruno, Odemir Martinez ^[3]

Afiliação do(s) autor(es)

^[1] Univ Fed Mato Grosso do Sul, BR-79907414 Ponta Pora, MS - Brazil

^[2] Univ Sao Paulo, Inst Math & Comp Sci, Sci Comp Grp, BR-05508 Sao Paulo - Brazil

^[3] Univ Sao Paulo, Sao Carlos Inst Phys, Sci Comp Grp, BR-05508 Sao Paulo - Brazil

• Tipo de documento: Artigo Científico
• Fonte: Neurocomputing; v. 153, p. 211-220, APR 4 2015.
• Citações Web of Science: 16

Resumo

In this paper, we propose a novel approach for dynamic texture representation based on complex networks. In the proposed approach, each pixel of the video is mapped into a node of the complex network. Initially, a regular complex network is obtained by connecting two nodes if the Euclidean distance between their related pixels is equal or less than a given radius. For each connection, a weight is defined by the difference of the pixel intensities. Given the regular complex network, a function is applied to remove connections whose weight is equal to or below a given threshold. Finally, a feature vector is obtained by calculating the spatial and temporal average degree for networks transformed by different values of threshold and radius. The number of connections of pixels from the same frame and from different frames, respectively, gives the spatial and temporal degrees. Experimental results using synthetic and real dynamic textures have demonstrated the effectiveness of the proposed approach. (C) 2014 Elsevier B.V. All rights reserved. (AU)

• Processo FAPESP: 11/02918-0 - Agentes, redes complexas e equações diferenciais parciais: integrando modelos matemáticos para análise de textura
• Beneficiário: Bruno Brandoli Machado
• Modalidade de apoio: Bolsas no Brasil - Doutorado
• Processo FAPESP: 11/01523-1 - Métodos de visão computacional aplicados à identificação e análise de plantas
• Beneficiário: Odemir Martinez Bruno
• Modalidade de apoio: Auxílio à Pesquisa - Regular