Precision Agriculture

, Volume 8, Issue 1, pp 95–110

Delineation of vine parcels by segmentation of high resolution remote sensed images

  • Jean Pierre Da Costa
  • Franck Michelet
  • Christian Germain
  • Olivier Lavialle
  • Gilbert  Grenier
Article

DOI: 10.1007/s11119-007-9031-3

Cite this article as:
Da Costa, J.P., Michelet, F., Germain, C. et al. Precision Agric (2007) 8: 95. doi:10.1007/s11119-007-9031-3

Abstract

Field delineation is an essential preliminary step for the design of management maps for grape production. In this paper, we propose a new algorithm for the segmentation of vine fields based on high-resolution remote sensed images. This algorithm takes into account the textural properties of vine images. It leads to the computation of a textural attribute on which a simple thresholding operation allows to discriminate between vine field and non-vine field pixels. The feasibility of the automatic delineation is illustrated on a range of vineyard images with various inter-row distances, grass covers, perspective distortions and side perturbations. In most cases it produces precise delineation of field borders while the parcel under consideration remains separate from the rest of the image.

Keywords

Remote sensingImage processingField segmentationRow cropVine

Copyright information

© Springer Science+Business Media, LLC 2007

Authors and Affiliations

  • Jean Pierre Da Costa
    • 1
    • 2
  • Franck Michelet
    • 1
  • Christian Germain
    • 1
    • 2
  • Olivier Lavialle
    • 1
    • 2
  • Gilbert  Grenier
    • 1
    • 2
  1. 1.Departement LAPS, IMSCNRS UMR 5218TalenceFrance
  2. 2.ENITA de BordeauxGradignan cedexFrance