Combined Segmentation of Lidar Point Cloud and Registered Images

By fusing with other sensory data, especially high resolution imagery, Lidar can be good source of information for DEM extraction and feature extraction. Nowadays airborne Lidar system vendors such as Leica and Toposys and others are providing systems (Leica ALS50II, ALS60, Toposys FALCON II) with integrated camera capturing 3D point cloud and high resolution images simultaneously. The full potential of the integrated system has to be explored yet. This paper presents an automatic segmentation method based on the fused data of point cloud and imagery. The method automatically partitions the scene by taking into account spectral, spatial and elevation information of pixels. The segmented regions contain multiple cues of object, which can further be used for feature extraction. The experimental result shows that the combined segmentation is useful for better DEM filtering and object classification than using single source of data (Lidar or imagery).