Identificação de Madeiras utilizando a Espectrometria no Infravermelho Próximo e Redes Neurais Artificiais

The identification of a species of tree becomes complex when it has available only your wood, which requires further analysis for their characterization. Using near infrared spectrometry is possible to obtain spectra with unique informations about the chemical composition of a sample timber. However, the interpretation of data obtained by the spectrometer is complex, making it difficult to identify characteristics specific to a particular species. In this paper, in order to speed up the identification process, we used a system based in Artificial Neural Networks for classification of four species of trees by analyzing the spectra of their timber. Three tests were performed to demonstrate the efficiency of recognition capability, the results obtained were encouraging since the neural net has proved to be flexible to noise and distortions, without requiring that the spectra were submit to prior statistical treatment or that were separated by groups relative to types of timber cut.

Artificial Neural Network; Levenberg-Marquardt heuristics; wood identification; near infrared spectroscopy