Joint VHR - LIDAR classification framework in urban areas using a priori knowledge and post processing shape optimization

In this paper we describe a joint methodology for exploiting multispectral and LIDAR data for the characterization of an urban area. The test site is the town of Uberlandia (Brazil). We first discuss the overall framework for 2D and 3D data fusion, and then introduce the approach investigated in this work. We then provide and discuss the mapping results obtained in our investigation. Finally, in order to increase the overall accuracy and enhance the extraction of single building/composite block shapes, a post-classification procedure is applied to the obtained map.