arXiv papers

Download open scientific papers. Check out the SCG's arXiv papers

Identificação de Plantas por Análise de Textura Foliar

André Ricardo Backes and Dalcimar Casanova and Odemir Martinez Bruno

Learning and Nonlinear Models, 9(1):84-90, 2011

The diversity of species in the plant kingdom become traditional models of taxonomy, in which the classification process is performed manually, a very difficult task. The present difficulties in this process imply in the existence of few studies of plant classification using mathematical and computational methods. Therefore, in order to contribute to the manuals methods of taxonomy already developed, this study presents a new computational method for identifying plant species through analysis of leaf texture. The method, called Fractal Multilevel, is based on the calculation of fractal dimension of binary images generated from the texture. The excellent results show how the computational methods for image analysis, texture analysis in particular, can help the task of identifying plant species.