A Knowledge-based approach to urban land use classification using AVIRIS imagery and LiDAR data

In this paper, the joint effect of hyperspectral and light detection and ranging (LiDAR) data for urban land use/ land cover (LULC) classification has been analyzed as combination of two data sources can result in better classification as compared to single data source. LULC classification of urban areas is a difficult task due to high spectral and spatial variability, especially with the use of single data source. The result of spectral angle mapper (SAM) classification, a supervised classification method, on hyperspectral imagery is compared with that of a knowledge based classification (KBC) combining LiDAR and hyperspectral data. Spectra from ASTER library was used as reference spectra for SAM classification while for Knowledge based classification nDSM derived from LiDAR data and indices derived from Hyperspectral data has been used. It was found that knowledge based classification had 7-8% more accuracy than SAM classification. Thus, it can be concluded that Knowledge based classification can be used as an efficient technique in this area