Urban land use information extraction using ALOS-2 PALSAR-2 data based on object-oriented method

Abstract

Both spatial texture details and polarization characterizations are contained in high resolution polarimetric synthetic aperture radar (PolSAR) data, which makes it possible to extract the land use information of urban areas. The object-oriented method has been proposed as a tool to process and analyze the high resolution PolSAR data. In this paper, a resolution-preserving PolSAR denoising method was used to preprocess the ALOS-2 PALSAR-2 data. Then a multi-scale segmentation technique based on spectral clustering was used to segment the preprocessed data. Finally, the buildings, roads, green space, and bare soil were extracted using the object-oriented method from the ALOS-2 PALSAR-2 data in some urban areas of Chengdu. Experiment results indicate that the oriented-objected method is effective for extracting the land use information of urban areas.