Author(s): BARMPOUTIS, P.; STATHAKI, T.; LLOYD, J.; MOURA, M. S. B. de; CARVALHO, H. F. de S.

Summary: To protect the value and potential of wood and forests, forest commissions, environmental agencies and scientists are focused on innovative solutions for environmental monitoring and management. To this end, a framework for tree classification that can contribute in the monitoring of wood resources, timber management and environmental protection is proposed in this paper. More specifically, we use LiDAR sensing technology and extract the skeleton of trees using the Fast Marching method. Then, we model them using linear dynamical systems taking advantage of the fact that each one of the tree species has been grown in a specific way and exhibits specific properties in the growing direction of it. This is achieved by dividing the tree skeletons into overlapping segments in the direction of z axes and creating higher-order patches consisting of trees? skeleton branching coordinates of each segment. Finally, due to the fact that the structure of tropical trees is complex, classification is performed introducing descriptors that take into account the combination of dynamic, appearance and noise parameters. For the evaluation of the proposed system, a dataset consisting of fifteen point clouds of common Caatinga tropical trees was created. Experimental results presented in this paper show the great application of the proposed methodology

Publication year: 2018

Types of publication: Paper in annals and proceedings

Unit: Embrapa Semi-arid Region

Keywords: Classificação de espécies, Floresta, Floresta Tropical, Monitoramento dos recursos madeireiros, Padronizar reconhecimento, Processamento de sinal multicanal, Remote sensing, Sensor remoto LiDAR, Sensoriamento Remoto, Sistemas dinâmicos lineares