Segmentation of noisy range images using the small-scale planarity of man-made vehicles
Michael C. Roggemann, James P. Mills, Steven K. Rogers, Matthew Kabrisky · Optical Engineering · 1991
Segmentation ofvehicles in natural backgrounds using absolute range imagery is considered. The segmentation technique exploits the small-scale planarity of man-made vehicles without explicit use of surface orientation information. A planarity test is used as the initial step in partitioning man-made vehicles from natural backgrounds. The planarity test is based on an estimate of the statistics of the absolute value of the error associated with fitting planes to small regions of the Cartesian coordinates of range image pixels. The planarity test is developed as a function of the standard deviation of the noise associated with range measurements. Heuristics used in conjunction with this algorithm are also presented. This algorithm has been evaluated on a database of 57 real range images, and results are presented. Application of the algorithm to two specific cases is shown in detail.