Land cover identification for finding hazelnut fields using WorldView-2 imagery

Remote sensing imagery is currently used as an efficient tool for agricultural management. Particularly, very high spatial resolution (less than 1m) enables extraction of permanent crops (including nut orchards) by visual interpretation or automated methods based on mainly textural features representing the regular plantation pattern. For accurate detection of orchards (hazelnuts in particular), this study proposes a rule-based classification utilizing multi-scale Gabor features and spectral values. Thanks to its very high spatial (0.5m) and spectral (8-bands) resolution, WorldView-2 imagery is primarily used. The classification accuracies, obtained with features extracted from WorldView-2 and Quickbird imagery, are compared for a study area in Turkey (major hazelnut producer in the world). In addition, supplementary value of the new 4 bands (coastal, yellow, red edge, and NIR2) in WorldView-2 imagery is discussed.