Splitting up panoramic range images into compact 2[1/2]D representations

This article presents a simple technique for splitting up a panoramic range image into a set of 2[1/2]D representations. The proposed technique consists of three stages. First, a spherical discretization map is generated. Second, main surface orientations are extracted together with their corresponding histogram of distances. Each one of these histograms is used to define the position of a projection plane as well as two associated clipping planes. Finally, data points bounded by clipping planes are mapped onto the corresponding projection plane defining a classical 2[1/2]D range image. This last stage—projection—is applied as many times as main orientations in the spherical discretization map. Experimental results with a panoramic range image are presented. VC 2006 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 16, 85–91, 2006; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ima.20069