Object-based Building Detection from Lidar Data and High Resolution Satellite Imagery

This paper presents a scheme for building detection from LIDAR data and high resolution satellite imagery. The proposed scheme comprises two major parts: (1) segmentation, and (2) classification. Spatial registration of LIDAR data and high resolution satellite images are performed as data pre-processing. It is done in such a way that two data sets are unified in the object coordinate system. Then, a region-based segmentation and object-based classification are integrated for building detection. In the segmentation, the LIDAR points are resampled to raster form. We, then, combine the elevation attribute from LIDAR data and radiometric attribute from orthoimages in the segmentation. The data with similar heights and spectral attributes are merged into a region. In the classification, we use the object-based classification to separate the building and non-building regions. The attributes considered in the classification include: (1) the elevation information from LIDAR data, (2) the spectral information from multispectral images, (3) the texture information from high spatial resolution image, (4) the roughness of LIDAR surface, and (5) the shape of regions. LIDAR data acquired by Leica ALS 40 and QuickBird satellite images were used in the validation.