Wood species identification using spectral reflectance

This paper describes an experiment in identifying the species of Douglas-fir, lodgepole pine, and spruce wood samples, based on their spectral-reflectance curves and quadratic discriminant analysis. Species predictions based on this method sometimes confused spruce with pine, but achieved an overall accuracy of 98.7 percent in separating Douglas-fir from spruce/pine. The paper suggests how comparable accuracies might be achieved in a system operating at production speeds in a mill environment.