Author: Burhan Baha Bilgilioğlu, Resul Çömert, Onur Yiğit, Ferhat Bedir
Publishing Date: 2019
Volume: 1 Issue: 1
The tea plant, which grows especially in the Eastern Black Sea region, constitutes a large part of the livelihood of the people living in this region. However, landslides that occur due to rainfall in the Eastern Black Sea Region cause continuous damage to the tea areas. In this respect, mapping of tea areas for temporal monitoring and observation is an important issue. In this study, it was aimed to determine tea gardens by using high resolution WorldView-2 satellite images. For this purpose, 8 km2 area was selected as a test area in Hopa district of Artvin province. Object-based image analysis was applied for the production of thematic maps of tea gardens. Multiresolution segmentation algorithm was used in the segmentation step. In the classification step, the K-nearest neighbor and random forest algorithm were used. When the obtained results were examined, it was found that random forest algorithm gave better results than K-nearest neighbor algorithm.
Key Words: Object Based Image, Analysis, Random Forest, K-NN, Tea Plant