Abstract
The morphological symptoms of phosphorus (P) deficiency in the leaves of
mini-cucumber plants at early stages of development have features similar to that of
early stage development in healthy plants. That similarity may lead to inappropriate
visual diagnostics of phosphorus deficiency in analyzed samples. Because the
differences in spectral properties of leaf tissues between phosphorus-deficient and
healthy plants can be demonstrated, the feasibility of using near-infrared (NIR)
spectroscopy for rapid and nondestructive diagnostics of phosphorus deficiency in
mini-cucumber plants was investigated. Leaf reflection spectra in the wavelength
range of 10 000-4000 cm<sup>−1</sup> were measured before the appearance of
morphological changes caused by phosphorus deficiency. Least-squares support vector
machine (LS-SVM), a method for recognizing patterns, was applied to identify
phosphorus-deficient plants. Parameters (γ, σ<sup>2</sup>) of LS-SVM were optimized
by cross-validation, and several conventional, two-class classification methods such
as linear discrimination analysis and K-nearest neighbors were also used
comparatively for identification. Identification rates in excess of 86% were
achieved with the LS-SVM model for both the training set and the prediction set. The
overall results indicated that NIR spectra combined with LS-SVM could be used
efficiently for pre-visual diagnostics of phosphorus deficiency in mini-cucumber
plants.
PDF Article
Cited By
You do not have subscription access to this journal. Cited by links are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.
Contact your librarian or system administrator
or
Login to access Optica Member Subscription