Automatic Classification of Compression Wood in Green Southern Yellow Pine


  • Jan Nyström
  • D. Earl Kline


Compression wood, color scanning, X-ray scanning, nondestructive evaluation, machine vision, image processing


Compression wood is a feature in softwoods that is undesired in sawn wood products due to its tendency to bend and crook as the moisture content changes. An automatic compression-wood detection method was developed and tested on southern yellow pine lumber in the green condition. Sixteen lumber specimens were scanned using both a color camera and an X-ray scanner. Color information was shown to have significant and consistent differences between compression wood and clear wood. However, X-ray information was found to contain large density variations in green lumber due to inconsistent moisture content that would mask density variations arising from compression wood. Therefore, it was concluded that X-ray information would not be useful in detecting compression wood in green southern yellow pine lumber. A multivariate regression model was developed based only on color information from one of the board samples. A nonlinear prediction model was produced by using the original color image data and expanded variables derived from the color images. The model based on one board sample was then applied on all boards. Classified images of the board surfaces were produced and compared to manually detected compression wood. An overall accuracy of 87% was observed in the classification of compression wood.


Andersson C., and F. Walter. 1995. Classification of compression wood using digital image analysis. Forest Prod. J. 45(11/12):87-92.nAnonymous. 1941. 1953. A simple device for detecting compression wood. US Forest Service FPL, Rep. 1390., Madison, WI.nBond, B. H., D. E. Kline, and P. A. Araman. 1998. Characterization of defects in lumber using color, shape, and density information. Pages 581-587 in Proc. International Conference on Multisource-Multisensor Information Fusion, CSREA Press.nBrunner C. C., A. G. Maristany, D. A. Butler, and J. W. Funck. 1993. Enhancing color-image data for wood-surface feature identification. Proc. 5th Int. Conf. on Scanning Technology & Process Control for the Wood Products Industry. Atlanta, GA.nConners R. W., D. E. Kline, P. A. Araman, and T. H. Drayer. 1997. Machine vision technology for the forest products industry. Computer 30(7):43-48.nDuda R., and P. Hart. 1973. Pattern classification and scene analysis. John Wiley & Sons, New York, NY.nGrundberg S. 1994. Scanning for internal defects in logs. Licenciate thesis, Luleå University of Technology, Luleå, Sweden.nHagman O. 1996. On reflections of wood. Ph.D. thesis, Luleå University of Technology, Luleå, Sweden.nLindgren O. 1992. Medical CT scanners for non-destructive wood density and moisture content measurements. Ph.D. thesis, Luleå University of Technology, Luleå, Sweden.nLindgren O., P. Geladi, and S. Wold. 1992. The kernel algorithm for PLS. J. Chemometrics 7:45-59.nNational Institutes of Health (NIH). NIH Image Version 1.62. April 1999.'> J. 1998. Detection of Picea abies compression wood with a multisensor system. Proc. 3rd International Seminar/Workshop on Scanning Technology and Image Processing on Wood. Luleå University of Technology. Technical Report 1998:27.nNystrom J., and O. Hagman 1999. Real-time spectral classification of compression wood in Picea abies. J. Wood Science 45(1):30-37.nPanshin A. J., and C. De Zeeuw. 1980. Textbook of wood technology. 4 ed. McGraw-Hill, New York, NY.nTimell T. E. 1986. Compression wood in gymnosperms. Springer Verlag., New York, NY.n






Research Contributions