A cylindrical neighborhood for multi-view range images processing. (March 2015)

• Record Type: Journal Article
• Authors: Shi, Bao-Quan; Liang, Jin
• Is Part Of: Optics and lasers in engineering. Volume 66(2015)
• Page Start: 240
• Page End: 248
• Publication Date: 2015-03
• Subjects: Multi-view range images -- Neighborhood -- Scanning -- Octree; Lasers in engineering -- Periodicals; Optical measurements -- Periodicals; Optics -- Periodicals; Lasers en ingénierie -- Périodiques; Mesures optiques -- Périodiques; Optique -- Périodiques; 621.36605
• DOI: 10.1016/j.optlaseng.2014.09.016
• Languages: English
• ISSNs: 0143-8166

Abstract

Abstract: The processing technologies of multi-view range images (MRIs) scanned by three-dimensional (3D) laser scanners, structured light scanners and laser radars (Ladars) are current research hotspot. Most of the processing such as denoising, integration and surface reconstruction are performed in local neighborhood. Thus, fast and precise query of the local neighborhoods of the 3D point data in MRIs is very important. Due to measurement accuracy and alignment error, there are gaps in overlapping regions and through theoretical analysis and experimental verification, we notice that the gaps which may result in loss of local neighborhood information have great influence on the neighborhood query whereas the previous methods ignore it. Thereby, a novel local neighborhood which is named cylindrical neighborhood (CYND) is proposed. Its searching procedures are described in detail. Compared with the previous methods, our CYND expands the searching scope along the directions of normal vectors of the 3D point data, hence eliminating the impact of gaps. The widely used and well-implemented ANN library was employed for a comparison study and the experimental results demonstrate the validity and superiority of our CYND. With extensive application of 3D optical scanning technologies, this novel neighborhood will have a wide application prospect. Highlights: A novel local neighborhood which is robust to gaps in the overlapping regions of MRIs is proposed. An Euclidean distance criterion and a normal vector criterion are developed to determine the neighborhood. A fast query procedure based on octree is developed to search neighbors of the 3D point data in the overlapping areas of MRIs.