3D Modelling of Cylindrical-Shaped Objects From Lidar Data - an Assessment Based on Theoretical Modelling and Experimental Data

Despite the increasing availability of measured laser scanning data and their widespread use, there is still the problem of rapid and correct numerical interpretation of results. This is due to the large number of observations that carry similar information. Therefore, it is necessary to extract from the results only the essential features of the modelled objects. Usually, it is based on a process using filtration, followed by simplification and generalization of redundant contents of datasets. This process must ensure the collection of new data without loss of information contained therein, the description accuracy of the key features of a measured object, as well as the uniqueness and comparability of results. In this paper, the authors extend the already extensive range of algorithms for the automatic or semiautomatic modelling of cylindrical objects, which have been measured using the laser scanning technology, with the one employing a concave hull or - in general - the alpha-shape (α-shape). The applicability of the proposed method was analysed using simulated data - generated analytically with the introduced and established systematic error with normal distribution - and using real results of the measurement of cylindrical-shaped objects. For real data, the obtained results were compared with the reference data. This made it possible to determine the validity of the proposed new method.