Comparison of neural and minimum distance classifiers in wood veneer defect identification

Abstract

The minimum distance classifier (MDC) is an example of a commonly used ‘conventional’ classifier. Whilst there has been a focus on using neural networks for the advantages that they offer, few researchers report direct comparison with conventional classifiers which typically have the advantage of being simpler. This paper provides such a comparison. The results show that the MDC does not perform as well as a neural network when applied to an industrial problem, namely that of identifying defects on wood veneer.

Keywords

minimum distance classifier multilayered feedforward neural network wood veneer inspection

References

Zurada

J. M.

Introduction to Artificial Neural Systems, 1992 (West Publishing Company).

Pham

D. T.

Liu

Neural Networks for Identification, Prediction and Control, 1997 (Springer-Verlag, London).

Haykin

Neural Networks: A Comprehensive Foundation, 1999 (Prentice-Hall, Englewood Cliffs, New Jersey).

Nelson

M. M.

Illingworth

W. T.

A Practical Guide to Neural Nets, 1991 (Addison-Wesley, Reading, Massachusetts).

Hertz

Krogh

Palmer

R. G.

Introduction to the Theory of Neural Computation: Lecture Notes, Vol. 1, 1991 (Addison-Wesley Reading, Massachusetts).

Taylor

T. G.

The Promise of Neural Networks, 1993 (Springer-Verlag, London).

Huber

H. A.

McMillin

C. W.

McKinney

J. P.

Lumber defect detection abilities of furniture rough mill employees. Forest Products J., 1985, 35 (11-12), 79–82.

Polzleitner

Schwingshakl

Real-time surface grading of profiled wooden boards. Ind. Metrology, 1992, 2, 283–298.

Packianather

M. S.

Drake

P. R.

Neural networks for classifying images of wood veneer - Part 2. Int. J. Advd Mfg Technol., 2000, 16(6), 424–433.

10.

Ruz

G. A.

Estévez

P. A.

Image segmentation using fuzzy min-max neural networks for wood defect detection. In Virtual International Conference on Intelligent Production Machines and Systems (IPROMS 2005), 4–15 July 2005; http://conference.iproms.org/.

11.

Rumelhart

D. E.

Hinton

G. E.

Williams

R. J.

Learning internal representations by error propagation. Parallel Distributed Processing: Explorations Microstructs Cognition, 1986, 1, 318–362.

12.

Pham

D. T.

Alcock

R. J.

Smart Inspection Systems, 2003 (Academic Press, London).

13.

Pham

D. T.

Alcock

R. J.

Automatic detection of defects on birch wood boards. Proc. Instn Mech. Engrs, Part E: J. Process Mechanical Engineering, 1996, 210, 45–52.

14.

Conners

R. W.

McMillin

C. W.

Lin

Vasquez-Espinosa

R. E.

Identifying and locating surface defects in wood: Part of an automated lumber processing system. IEEE Trans. Pattern Analysis Mach. Intell., 1983, 5 (6), 573–583.

15.

Koivo

A. J.

Kim

C. W.

Classification of surface defects on wood boards. In Proceedings of the IEEE International Conference on Systems, Man and Cybernetics, Atlanta, Georgia, USA, 1986, pp. 1431–1436 (IEEE, New York).

16.

Sobey

P. J.

Automated optical grading of timber. Proc. SPIE, 1990, 1379, 168–179.

17.

Lappalainen

Alcock

R. J.

Wani

M. A.

Plywood feature definition and extraction. Report 3.1.2, QUAINT, BRITE/EURAM project 5560, Intelligent Systems Laboratory, School of Engineering, Cardiff University, 1994.

18.

Packianather

M. S.

Drake

P. R.

Rowlands

H. R.

Optimizing the parameters of multilayered feedforward neural networks through Taguchi design of experiments. Qual. Reliability Engng Int., 2000, 16 (6), 461–173.

19.

Packianather

M. S.

Drake

P. R.

Modelling neural network performance through response surface methodology for classifying wood veneer defects. Proc. Instn Mech. Engrs, Part B: J. Engineering Manufacture, 2004, 218, 459–466.