BUILDING EXTRACTION USING SURFACE MODEL CLASSIFICATION

In many applications like urban planning and environmental simulation, the major solution is building extraction, which can be performed using different airborne or space-borne data or an appropriate fusion of them. This paper presents an automatic building recognition technique using fusion of LIDAR data and multispectral imagery. To this end, a rule-based classification method is considered in order to extract buildings from input data which are DSM, DTM extracted from DSM and an optical Image. To achieve better accuracy classification is performed in both pixel and object level. Accordingly, a user-friendly MATLAB toolbox is provided for both classification and evaluation procedures. It is experimentally shown that the proposed algorithm can successfully detect urban residential buildings, when assessed in terms of different quantitative criteria and visual inspection.