期刊名称：Archiwum Fotogrametrii, Kartografii i Teledetekcji
印刷版ISSN：2083-2214
出版年度：2006
卷号：16
语种：English
出版社：Main Board of Association of Polish Surveyors
摘要：Within recent years,airborne laser scanning has become the leading technology of acquiring discrete geometrical information about ground surface and objects existing on it. The main issue in data processing is the identification and classification of points belonging to the proper surfaces. Thereafter,these points constitute the basis for surface modelling. The problem of classification concerns roofs of buildings and the topographical ground surface. The latter one is the subject of the research in the presented work. The form of an automatic elimination of points not belonging to the modeled surface is called a filtration. Based upon a literature review concerning airborne laser scanning data filtration,the analysis and experience of authors in data processing,the following filtration requirements can be formulated: - if possible,the filtration process should be carried out on the original data, - the algorithm should have the property of suitable conformity to the local terrain structures, - the algorithm should take into consideration additional information a-priori, - due to the very large size of laser scanning data sets,in the order of 106 points,the complexity of the filtration algorithm is important. In the context of above mentioned requirements,the method of moving polynomial surface in the process of airborne laser scanning data filtration was adapted in this work. The small rank polynomial surface was locally fit to the measured data in the iteration process. Parameters of the surfaces were calculated based upon M-estimators of the robust estimation method. In the estimation process,the distance inverse function as the weighting function and the asymmetrical damping function were used. The filtration algorithm was applied using the hierarchical method. The algorithm tests were executed on the real data captured by ScaLARS system. In order to evaluate the correctness of the algorithm,hand points classification using air photos,calibrated in the same coordinate system,was carried out. The terrain points identification accuracy was evaluated on the level of about 2 %. A discussion of results is included and the advantages and defects of the presented algorithm are discussed in the paper. The proposed algorithm is relatively simple,and is based upon the original data and,due to free choice of weighting and damping functions parameters,has properties of suitable approximation of the local terrain structures.
其他摘要：Lotniczy skaning laserowy dostarcza punktowej informacji geometrycznej o fizycznej powierzchni ziemi i znajdujących się na niej obiektach. W procesie opracowania danych centralnym zagadnieniem jest identyfikacja punktów naleŜących do odpowiednich powierzc
关键词：laser scanning;filtering;moving polynomial surface;robust estimation;weighting function
其他关键词：skaning laserowy;filtracja;wielomianowa powierzchnia ruchoma;estymacja odporna;funkcja wagowa;funkcja tłumienia
