A NEW APPROACH FOR SEGMENTATION-BASED TEXTURING OF LASER SCANNING DATA

Abstract

Abstract. In recent years, laser scanning systems have been recognized as a fast and accurate technology for the acquisition of high density spatial data. The advent of these systems has reduced the cost and increased the availability of accurate 3D data for mapping, modelling, and monitoring applications. The original laser scanning data does not explicitly provide meaningful information about the characteristics of the scanned surfaces. Therefore, reliable processing procedures are applied for information extraction from these datasets. However, the derived surfaces through laser scanning data processing cannot be effectively interpreted due to the lack of spectral information. To resolve this problem, a new texturing procedure is introduced in this paper to improve the interpretability of laser scanning-derived surfaces using spectral information from overlapping imagery. In this texturing approach, individual planar regions, derived through a laser scanning data segmentation procedure, are textured using the available imagery. This texturing approach, which aims to overcome the computational inefficiency of the previously-developed point-based texturing techniques, is implemented in three steps. In the first step, the visibility of the segmented planar regions in the available imagery is checked and a list of appropriate images for texturing each planar region is established. An occlusion detection procedure is then performed to identify the parts of the segmented regions which are occluding/being occluded by other regions in the field of view of the utilized images. In the second step, visible parts of planar regions are decomposed into parts which should be textured using individual images. Finally, a rendering procedure is performed to texture these parts using available images. Experimental results from real laser scanning dataset and overlapping imagery demonstrate the feasibility of the proposed approach for texturing laser scanning-derived surfaces using images.