Automatic discrimination of grapevine (Vitis vinifera L.) clones using leaf hyperspectral imaging and partial least squares

  1. Fernandes, A.M. 1
  2. Melo-Pinto, P. 1
  3. Millan, B. 2
  4. Tardaguila, J. 2
  5. Diago, M.P. 2
  1. 1 Universidade de Trás-os-Montes e Alto Douro
    info

    Universidade de Trás-os-Montes e Alto Douro

    Vila Real, Portugal

    ROR https://ror.org/03qc8vh97

  2. 2 Instituto de Ciencias de la Vid y del Vino
    info

    Instituto de Ciencias de la Vid y del Vino

    Logroño, España

    ROR https://ror.org/01rm2sw78

  3. 3 Catholic University of the Sacred Heart
    info

    Catholic University of the Sacred Heart

    Milán, Italia

    ROR https://ror.org/03h7r5v07

Mostrar afiliaciones +
Revista:
Journal of Agricultural Science

ISSN: 0021-8596

Año de publicación: 2015

Volumen: 153

Número: 3

Páginas: 455-465

Tipo: Artículo

DOI: 10.1017/S0021859614000252 SCOPUS: 2-s2.0-84924857324 WoS: WOS:000351414900007 GOOGLE SCHOLAR

Otras publicaciones en: Journal of Agricultural Science

Repositorio institucional: lock_openAcceso abierto Postprint lock_openAcceso abierto Editor

Resumen

A worldwide innovative method to discriminate grapevine clones is presented. It is an alternative to ampelography, isozyme and DNA analysis. The spectra and their first and second derivatives of 201 bands in the visible and near-infrared wavelength range between 634 and 759 nm were used as inputs to a classifier created using partial least squares. The spectra were acquired in the laboratory for the adaxial side of the apical part of the main lobe of fully hydrated grapevine leaves. The classifier created allowed the separation of 100 leaves of the Cabernet Sauvignon (Vitis vinifera L.) variety into four clones, namely CS 15, CS 169, CS 685 and CS R5, comprising 25 leaves each. The percentages of leaves correctly classified for these clones were 98·2, 99·2, 100 and 97·8%, respectively, when the classifier input was the second derivative of the normalized spectra. These percentages were determined by Monte-Carlo cross-validation. With the new method proposed, each leaf of a given variety can be classified in a few seconds according to its clone in an environmentally friendly way. Copyright © Cambridge University Press 2014.