Determining bark content in wood-chips for pulping with computerized image analysis

This thesis examines the possibility to use computerized image analysis for quality control of wood-chips used for pulping. The quality control in this case means controlling the amount of bark in the chips, since bark can create damages in the finished product. At present the chips quality at Stora Enso Packaging Boards, Skoghall Mill is controlled with samples where the amount of bark is evaluated. The evaluation is done by sorting out the bark in the sample and calculating the weight percentage. To examine if this controll is possible to do with computerized image analysis two different sets of images were used. Image series 1 was used for developing and evaluating the method and image series 2 was used only for evaluation. The results from image series 1 are good and shows that it is possible to separate the bark from the rest of the pieces. In image series 2 the results are not quite as good. Not all bark pieces are found, and some other pieces are instead identified as bark. The reason for this is that the image material for developing the algorithm was not large enough. Continuing to develop the method with more images from different sets of wood-chips would probably give even better results.