Abstract
This study aims to develop agricultural land use mapping technique for detection and classification of unused farmlands. Common techniques currently in use utilize the pixel as base unit for farmland mapping; this works well in big area plots (Example: China, USA), but looses efficiency in small plots common to Japanese farmlands. The developed technique targets this problem by detecting and classifying plot divisions based on remote sensing data. Image segmentation of aerial photographs generated various polygons as farmland plot divisions. These polygons were classified according to space analysis of satellite images. The results proved very useful allowing the production of unused farmland maps.
