El profesor Jesús Alpuente Hermosilla, ha defendido su tesis doctoral en la Universidad de Alcalá, en un tema muy forestal que es poner las bases científicas para clasificar de forma automática la especie a la que pertenece una determinada muestra de madera, sin más requisitos que disponer de un escáner convencional o cámara de fotos con las que obtener una imagen digital de la superficie de la muestra y de tener previamente digitalizado muestras patrón de las distintas especies de madera.
Professor Jesús Alpuente Hermosilla has defended his doctoral thesis at the University of Alcalá, in a forest theme, which is to scientific basis for automatically classify the species ti which a given sample timber belongs, without any requeriment to have s scanner conventional or camera with which obtain a digitized surface and having previously digitized standard samples of different wood species image.
The methodology is based on the use of pattern matching tecniques based on Decision Theory: From the scanned sample under study and with the help of public domain software packages (ImageJ, and Weka), using a novel approach inspired by Differential Contrast Interference Microscopy, pseudo-3D images and periodic pattern of wood surface are developed. (see picture). Theses developed images distinct linnear classifiers based on decision trees, with different training methods.
The results prove that this methodology provides a high degree of reliability, in many cases exceeding 95% correct.
