VECTOR-RASTER DATA FUSION FOR OBJECT EXTRACTION
This paper proposes an approach to identify and extract automatically objects from a combination of a digital surface model (DSM) created from aerial laser scanning data and high-resolution satellite images. The approach consists of two steps. The first step is a MapReduce process where neighboring points in a DSM are mapped into cubes. The second step uses an algorithm to extract adjacent cubes. According to this algorithm, all adjacent cubes belong to the same object. Near the ground level, the algorithm extracts only cubes whose coordinates match with the coordinates of the corresponding outline pixels of satellite images. Finally, an evaluation study was done for a port in Zeebruges, Belgium to demonstrate the applicability of the approach. The proposed approach is notable not only for its big data context but its usage of vector and raster data.
building extraction, MapReduce, big data, LiDAR, high-resolution satellite images, digital surface model, aerial laser scanning.