Spectroscopic determination of health levels of Coast Live Oak (Quercus agrifolia) Leaves
R. Pu1, Q. Chen2, D. Graham-Squire2 and P. Gong2
A total of 153 reflectance spectra (covering 350 - 2500 nm) from coast live oak (Quercus agrifolia) leaves were measured in the laboratory with a spectrometer FieldSpec®Pro FR for "Sudden Oak death (SOD)" monitoring. It is well known that the spectroscopic determination of health status of the coast live oak leaves is very difficult due to subtle spectral differentiation between different health levels. In this study, two spectroscopic classification algorithms, cross correlogram spectral matching (CCSM) and penalized discriminant analysis (PDA), are applied to identify the two health levels of oak leaves: healthy and infected. CCSM is practiced by calculating the cross correlation at different match positions between a test spectrum and a reference spectrum. A test spectrum with a higher cross correlation will have a perfect matching to a reference spectrum, which leads to the test spectrum classified to the reference spectrum. PDA is a penalized version of Fisher's linear discriminant analysis (LDA) and can considerably improve upon LDA when it is used in classification of hyperspectral data. In this experiment, the 153 spectra (54 for healthy and 99 for infected) was randomly divided into three non-overlapping groups, and among them two groups were used as training samples and the remaining one as test samples (repeating the selection three times for obtaining three non-overlapping groups of test samples) for testing the two classification algorithms. Experimental results indicate that the PDA algorithm has produced a slightly higher classification accuracy (68.0%) than that (65.2%) by CCSM for identifying the two health levels.
1Center for Assessment & Monitoring of Forest & Environmental Resources (CAMFER), 145 Mulford Hall, University of California, Berkeley, Ca 94720-3114, USA; (510) 642-1351; Fax: (510) 643-5098; rpu@nature.berkeley.edu
2Center for Assessment & Monitoring of Forest & Environmental Resources (CAMFER), 145 Mulford Hall, University of California, Berkeley, Ca 94720-3114, USA
|
