Fourier Bessel Characterization of Polished Metal Surfaces
The amount of information available from metallographic cross sections about material properties remains only subjectively accessible via trained metallurgical experts. Image analysis has not fulfilled the promise to quantify, standardize, or transmit this subjective information. Our solution is to describe different alloys with Luerkens-Beddow style Bessel Fourier descriptors. When AISI-SAE 1035, 1045, and 1060 steels undergo spheroidizing or normalizing heat treatments, the relative amounts and morphology of pearlite, bainite, and other phases, which determine material properties, change even more. The Bessel Fourier descriptors enable accurate classification (up to 93% accuracy) according to heat treatment history and alloy composition, indicating their superior ability to describe metallographic surfaces.
KeywordsHeat Treatment Classification Accuracy Discrimination Accuracy Stepwise Discriminant Analysis Discrimination Function
Unable to display preview. Download preview PDF.
- 2.SAS, “SAS/STAT User’s Guide: Release 6.03 Edition”, SAS Institute, Cary, North Carolina.Google Scholar
- 3.Press, W.H. et. al. Numerical Recipes Cambridge Univ. Press, 1986 ISBN 0 521 30811 9.Google Scholar
- 4.Prisbrey, Keith, “Comparison of Bessel Fourier Coefficients to Gray Level Histograms as Pattern Recognition Features”, presented at Fine Particle Society Meeting, San Jose, CA, August, 1988.Google Scholar
- 5.Luerkens, D. W. and Beddow, Keith, “Morphological Analysis”, Short Course presented at the Fine Particle Society Meeting, San Jose, CA, August, 1988.Google Scholar