Percolation Features: An approach for evaluating fractal properties in colour images[Formula presented]

• Autor(a) principal: Roberto, Guilherme Freire
• Data de Publicação: 2022
• Outros Autores: Neves, Leandro Alves [UNESP], da Costa Longo, Leonardo Henrique [UNESP], Rozendo, Guilherme Botazzo [UNESP], Tosta, Thaína Aparecida Azevedo, de Faria, Paulo Rogério, Martins, Alessandro Santana, do Nascimento, Marcelo Zanchetta
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Repositório Institucional da UNESP
• Texto Completo: http://dx.doi.org/10.1016/j.simpa.2022.100387; http://hdl.handle.net/11449/241480
• topic: Feature extraction; Fractal geometry; Pattern recognition; Percolation

Percolation is a physical phenomenon that describes the transport of fluids through porous media. We present a new approach for extracting local and global percolation features from RGB images. The proposed method is based on multiscale and multidimensional approaches for extracting other fractal measures from histology images. Our method is able to obtain three percolation properties: number of clusters, occurrence of percolation, and size of the largest cluster. This approach aimed to provide an alternative to fractal dimension and lacunarity as ways of evaluating fractal properties in an image.