A THEMATIC MAP FOR UPM, ENGINEERING CAMPUS PRODUCED BY OBJECT-BASED CLASSIFICATION METHOD

摘要

Thematic maps generated by image classification process are useful for numerous geospatial applications such as urban planning, natural disasters, site selection, and change detection. This study proposed a classification workflow based on object-based analysis forproducing thematic maps from Worldview-2 satellite data. The image was segmented by multiresolution segmentation algorithm in eCognition software and image objects were classified by support vector machine. Results indicate that the propose method is suitable for thematic map creation. The overall accuracy achieved in this study was 89.9% and the kappa was 0.86. Findings indicate the usefulness of the object-based image analysis for very high-resolution image processing and producing thematic maps at small and medium scales. Future works should focus on optimization of parameters required by algorithms and improving the segmentation results.