Premium
The application of some image‐analysis techniques to recognition of soil micromorphological features
Author(s) -
TERRIBILE F.,
FITZPATRICK E.A.
Publication year - 1995
Publication title -
european journal of soil science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.244
H-Index - 111
eISSN - 1365-2389
pISSN - 1351-0754
DOI - 10.1111/j.1365-2389.1995.tb01810.x
Subject(s) - principal component analysis , brightness , point (geometry) , computer science , pattern recognition (psychology) , transformation (genetics) , feature (linguistics) , anisotropy , perspective (graphical) , artificial intelligence , matrix (chemical analysis) , image (mathematics) , remote sensing , computer vision , mathematics , geology , optics , materials science , geometry , physics , chemistry , biochemistry , linguistics , philosophy , composite material , gene
Summary A number of image analysis techniques, largely drawn from remote sensing, have been applied to soil thin sections in an attempt (i) to overcome some of the limitations of previous approaches and (ii) to identify some complex soil features. Using different wavelengths and light polarizations, microscopic images of thin sections obtained from different horizons were digitized. It was then possible to carry out a series of experiments including geometric corrections, creation of multilayer images, principal component transformation, classification procedures, morphological analysis and modelling of bi‐dimensional optical anisotropy of the soil matrix. As a special procedure, principal component images were included in the multilayer images, thereby improving the classification on which a morphological analysis was conducted. The reliability of the procedure was tested against point counting and was found to be successful. The overall procedure allowed the combination of brightness and shape classifications. Bi‐dimensional matrix optical anisotropy was detected at 256 levels and plotted as a three‐dimensional perspective view, thus creating a new way of studying this feature.