文章基本信息

标题：APPLICATION OF PARALLEL COMPUTING FOR CLASSIFICATION OF OVERLAPPING POINTS
其他标题：ASTOSOWANIE OBLICZEŃ RÓWNOLEGŁYCH DO KLASYFIKACJI PUNKTÓW OVERLAP
本地全文：下载
作者：Ryszard Bratuś ; Paweł Musialik ; Piotr Pióro 等
期刊名称：Archiwum Fotogrametrii, Kartografii i Teledetekcji
印刷版ISSN：2083-2214
出版年度：2017
卷号：29
DOI：10.14681/afkit.2017.002
语种：English
出版社：Main Board of Association of Polish Surveyors
摘要：The paper presents innovative methods of solving important technological problem: the classification of LiDAR points located in the overlapping area between two parallel scan strips. The presented approach is based on an efficient method of parallel computation using graphic processors, allowing to apply more sophisticated algorithms for data analysis and processing. The tests of the algorithms were executed in order to verify correctness of the assumption that the innovative solutions presented in the paper might increase the efficiency and correctness of the data, referred to well known and popular technological solutions. The suggested computational innovations are applied to increase the quality of the LiDAR data acquired by light airplanes and gyrocopters. Two approaches to increase the quality of classification of overlapping points have bee,proposed. The first process is cutting-off the points of the strip borders strictly according to defined angle measured from vertical direction. The second process is dissolving of the points to get the regular density of the result point cloud. The title issue is the classification of overlapping points according to the angle of incidence to the terrain and other objects. The normal vectors calculation for each of the scan points is necessary for the analysis. Such solution increases the quality of overlaps classification and guarantees its high efficiency thanks to the parallel computation. In conclusion,during the research three innovative approaches were tested and reviewed against commonly used methods. Parallel computation can improve quality and reduce time of processing for overlap classification problem was confirmed. .
其他摘要：Publikacja omawia nowatorskie metody rozwiązania ważnego technologicznie zagadnienia,jakim jest klasyfikacji punktów overlap,czyli punktów w pasie podwójnego pokrycia pomiędzy sąsiednimi szeregami skanowania. Prezentowane podejście oparte jest na wydajnej
关键词：overlapping points;parallel computing;points classification;scan angle
其他关键词：punkty typu overlap;przetwarzanie równoległe;klasyfikacja punktów;kąt skanowania