AI Summary & Key Takeaways
AI InsightOur results indicated that, decision-tree oriented GEOBIA classification provided sufficient results for determination of olive trees with 90 percent classification accuracy and differentiating them from non- vegetated areas and annual crops.
📌 3 Key Takeaways
Abstract
Identification of fruit trees and determination of their spatial distribution is an important task for several agricultural activities including fruit yield estimation, irrigation planning, disease management and supporting agricultural policies. This research aims to determine spatial distribution of olive trees at parcel level by using geographic object based image analysis (GEOBIA) and very high resolution satellite images. A pilot area located in the Aegean region of Turkey was selected to conduct research considering the massive amount of olive production within the area. GEOBIA based decision-tree classification was applied to accurately map perennial crop parcel boundaries. After applying multi-resolution segmentation to create image objects, thresholds determined from spectral properties of image objects were integrated into the decision tree to ensure accurate mapping of olive trees. Accuracy assessment was conducted by comparing a highly accurate parcel database with classification results and efficiency of parcel identification and areal information derivation were evaluated. Our results indicated that, decision-tree oriented GEOBIA classification provided sufficient results for determination of olive trees with 90 percent classification accuracy and differentiating them from non- vegetated areas and annual crops. Area estimation and parcel detection performances of the method were also acceptable by providing 0.11 and 0.08 relative errors respectively.