Automatic quality sorting of Picea abies logs with a gamma ray log scanner

Abstract

A quality index for quality sorting of Norway spruce (Picea abies) has been developed. The quality index expresses the expected mean quality of the centerboards from the log. This quality index is based on property variables obtained when using a gamma ray scanner. A multivariate calibration method, PLS (partial least squares models in latent variables), was used. A training set of 2 000 graded centerboards coming from 1 000 selected spruce logs was used to create the model between measured property variables and the expected quality output. The model is tested with crossvalidation within the training set, and expected precision of aim is 65% overall correct classification and a 70% correct classification for dimensions 150 to 225 mm. It can be utilized when sorting logs automatically with respect to the quality of the center boards.