Multispectral images were collected by an unmanned aerial vehicle over a commercial coffee plantation duringthe 2002 harvest season. Selected scenes were georegistered to a base map and a mosaic of the study area was created. Imagesegmentation was performed to identify and mask soil, shadow, and cloud pixels. The remaining pixels, representing sunlitcanopy, were assumed to be a mixture of four components: green leaf, underripe fruit, ripe fruit, and overripe fruit. Field andlaboratory instruments were used to measure the reflectance spectrum of each component. Based on these spectra, a ripenessindex was developed for the airborne imagery that involved computing the per-pixel ratio of digital counts in spectralchannels centered at 580 and 660 nm. Results were aggregated on a per-field basis. Mean ripeness index per field wassignificantly correlated with ground-based counts recorded by the grower, and to eventual harvest date. The results suggestthat remote sensing methods may provide an alternative, more spatially comprehensive method for monitoring ripeness statusand evaluating harvest readiness of this high-value agricultural commodity.
No takes yet. Share an insight, caveat, or question.
Johnson et al. (2004) studied this question.