Identification of Log Characteristics in Computed Tomography Images Using Back-Propagation Neural Networks with the Resilient Back-Propagation Training Algorithm and Textural Analysis: Preliminary Results
Keywords:
Black spruce, sugar maple, log characteristics, computed tomography (CT) images, artificial neural networksAbstract
This research addressed the feasibility of identifying internal log characteristics in computed tomography (CT) images of sugar maple and black spruce logs by means of back-propagation (BP) neural networks with a resilient BP training algorithm. Five CT images were randomly sampled from each log. Three of the images were used to develop the corresponding classifier, and the remaining two images were used for validation. The image features that were used in the classifier were gray-level values, textual, and distance features. The important part of the classifier topology, ie the hidden node number, was determined based on the performance indicators: overall accuracy, mean square error, training iteration number, and training time. For the training images, the classifiers produced class accuracies for heartwood, sapwood, bark, and knots of 99.3, 100, 96.7, and 97.9%, respectively, for the sugar maple log; and 99.7, 95.3, 98.4, and 93.2%, respectively, for the black spruce log. Overall accuracies were 98.5% for sugar maple and 96.6% for black spruce, respectively. High overall accuracies were also achieved with the validation images of both species. The results also suggest that using textural information as the inputs can improve the classification accuracy. Moreover, the resilient BP training algorithm made BP artificial neural networks converge faster compared with the steepest gradient descent with momentum algorithm. This study indicates that the developed BP neural networks may be applicable to identify the internal log characteristics in the CT images of sugar maple and black spruce logs.References
Bhandarkar SB, Faust TD, Tang MJ (1999) Catalog: A system for detection and rendering of internal log defects using computer tomography. Mach Vis Appl 11(4):171-190.nBucur V (2003) Nondestructive characterization and imaging of wood. Springer Series in Wood Science. Springer-Verlag, Berlin, Germany. 337 pp.nCongalton R (1991) A review of assessing the accuracy of classifications of remotely sensed data. Remote Sens Environ 37(1):35-46.nEinspahr DW, Harder M (1975) Hardwood bark properties important to the manufacture of fiber products. IPC Techn Pap Ser No. 11:11-28.nForest Products Laboratory (1999) Wood handbook. University Press of the Pacific, Madison, WI. 464 pp.nFreeman JA, Skapura DM (1991) Neural networks algorithms, applications and programming techniques. Addison-Wesley Publishing Company. 401 pp.nGhosh MN, Sharma D (1963) Power of Tukey's test for non-additivity. JR Stat Soc [Ser A] 25(1):213-219.nGrundberg S (1994) Scanning for internal defects in logs. Licentiate Thesis, Luleå University of Technology. 121 pp.nHaralick RM, Shanmugam K, Dinstein I (1973) Texture characteristics for image classification. IEEE T Syst Man Cyb 3(6):610-621.nHou ZQ, Zhang SY, Hu LJ (2005) Determining wood density and its 3-D distribution in a log using computed tomography scanning technique. Pp 427-423 in Proc 14th International Symposium on Non-destructive Testing of Wood, May 2-4, 2005, Shaker Verlag, Germany.nHounsfield GN (1980) Computed medical imaging. Forest Sci 210(4465):22-28.nLamb FM, Marden RM (1970) Variation in density of sugar maple sapwood and heartwood. USDA For Prod Lab Res Note NC-90.nLemieux H, Beaudoin M, Zhang SY, Grondin F (2002) Improving structural lumber quality in a sample of Picea mariana logs sawn according to the knots. Wood Fiber Sci 34(2):266-275.nLi P, He J, Abbott L, Schmoldt DL (1996) Labeling defects in CT images of hardwood logs with species-dependent and species-independent classifiers. Pp 113-126 in Proc IAPR TC-8 Workshop on Machine Perception Applications, September 2-4, 1996, Technical University Graz, Graz, Austria.nMas JF, Flores JJ (2008) The application of artificial neural networks to the analysis of remotely sensed data. Int J Remote Sens 29(3):617-663.nNordmark U (2002) Knot identification from CT images of young Pinus sylvestris sawlogs using artificial neural networks. Scand J For Res 17(1):72-78.nOcceña LG (1991) Computer integrated manufacturing issues related to the hardwood log sawmill. J Forest Eng 3(1):39-45.nPang S, Wiberg P (1998) Model predicted and CT scanned moisture distributed in a Pinus radiata board during drying. Holz Roh Werkst 56(1):9-14.nPanshin AJ, Zeeuw CD (1971) Textbook of wood technology: Volume 1. 3rd ed. McGraw-Hill Book Company, New York, NY. 705 pp.nRichards DB (1977) Value yield from simulated hardwood log sawing. Forest Prod J 27(12):47-50.nRichards DB (1980) Lumber value for computerized simulation of hardwood log sawing. USDA For Prod Lab Res Pap FPL-356, Madison, WI.nRiedmiller M, Braun H (1993) A direct adaptive method for faster back-propagation learning: the RPROP algorithm. Pp 828-845 in Proc IEEE International Conference on Neural Network, March 28-April 1, 1993, San Francisco, CA.nRojas G, Condal A, Beauregard R, Verret D, Hernandez RE (2006) Identification of internal defect of sugar maple logs from CT images using supervised classification methods. Holz Roh Werkst 64(4):295-303.nRojas G, Hernández RE, Condal A, Verret D, Beauregard R (2005) Exploration of physical properties of internal characteristics of sugar maple logs and relationships with CT images. Wood Fiber Sci 37(4):591-604.nRosin PL, Fierens F (1995) Improving neural network generalization. Pp 1255-1257 in Proc Internation Geoscience and Remote Sensing Symposium-IGARSS, July 10-14, 1995, Firenze, Italy.nSchmoldt DL, He J, Abbott AL (2000) Automated labeling of log features in CT imagery of multiple hardwood species. Wood Fiber Sci 32(3):287-300.nSjoberg J (1995) Nonlinear black box modeling in system identification: A unified overview. Automatica 31(12): 1691-1724.nWei Q, Leblon B, Chui YH, Zhang SY (2008) Identification of selected log characteristics from computed tomography images of sugar maple logs using maximum likelihood classifier and textural analysis. Holzforschung 62(4):441-447.nZhu DP, Conners RW, Schmoldt DL, Araman PA (1996) A prototype vision system for analyzing CT imagery of hardwood logs. IEEE T Syst Man Cyb 26(4):522-532.n
Downloads
Published
Issue
Section
License
The copyright of an article published in Wood and Fiber Science is transferred to the Society of Wood Science and Technology (for U. S. Government employees: to the extent transferable), effective if and when the article is accepted for publication. This transfer grants the Society of Wood Science and Technology permission to republish all or any part of the article in any form, e.g., reprints for sale, microfiche, proceedings, etc. However, the authors reserve the following as set forth in the Copyright Law:
1. All proprietary rights other than copyright, such as patent rights.
2. The right to grant or refuse permission to third parties to republish all or part of the article or translations thereof. In the case of whole articles, such third parties must obtain Society of Wood Science and Technology written permission as well. However, the Society may grant rights with respect to Journal issues as a whole.
3. The right to use all or part of this article in future works of their own, such as lectures, press releases, reviews, text books, or reprint books.
