Improving pesticide spray application in vineyards by automated analysis of the foliage distribution pattern in the leaf wall

A novel color-based recording and evaluation technique is presented which allows to determine the spatial distribution of foliage along approximately planar vine rows with high accuracy. The approach uses optical flow tracking to construct a composite from an image sequence and user-trainable bayesian color classification to segment foilage from a distinctively colored canvas carried along behind the row under evaluation. The gained information furthers insight into the seasonal growth patterns of commercial plantations and also allows to validate the performance of new-generation, selective spraying pesticide sprayers.