Segmentation of 3D points from range camera data using scanlines

This paper focuses on the segmentation of 3D points from range camera data. Range cameras can capture 3D information of a scene by sending out infrared light and then measuring the reflections. A segmentation algorithm for the detection of planar surfaces is presented which is robust against erroneous data from the range cameras. It uses a grid of horizontal and vertical scanlines as intermediate primitives for the final planar region segmentation. By using the scanlines, the task of planar segmentation can be separated into two tasks which are less complex: line segmentation and line combination. The usage of higher primitives improves the segmentation time compared to the direct operation on the 3D point data. The algorithm is described in detail and evaluated on a test scene.