ARTIFICIAL MODEL IN THE ASSESSMENT OF THE ALGORITHM OF OBJECTS RECORDED BY LASER SCANNING SHAPE DETECTION (ALS/TLS)

• Published in: International Multidisciplinary Scientific GeoConference SGEM Vol. 2; no. 1; pp. 995 - 1002
• Main Authors: Janowski, Artur, Nierebinski, Piotr, Szulwic, Jakub
• Format: Conference Proceeding
• Language: English
• Published: Sofia Surveying Geology & Mining Ecology Management (SGEM) 01.01.2015
• Subjects: Airborne lasers; Algorithms; Automation; Data analysis; Data processing; Edge detection; Effectiveness; Empirical analysis; Laser applications; Lasers; Lidar; Parameter identification; Segmentation; Software; Spatial data; Terrain models; Verification
• ISSN: 1314-2704

Brief description of the study and used methods. Brief description of the study and used As part of the preparatory work aimed to create the application solution allowing for the automation of searching objects in data, obtained in the scanning process using ALS (Airborne Laser Scanning) or TLS (Terrestrial Laser Scanning), the authors prepared a artificial (synthetic, theoretical) model of space, used for the verification of operation of algorithm based on the images analysis. The research described in the paper was based on synthetically generated model of points cloud which gave bases to continue works on real data sets obtained from the registration with the use of airborne scanner (LiDAR). Authors attempt to point out the importance (and essence) of analyses using a synthetic model in the analysis of authors algorithms (including those based on the use of the integration of known algorithms), applicable to the surveying works aimed to identify objects on results of laser scanning or filtration in order to obtain the numerical terrain model. In modern IT publications, is possible to find solutions that relate to authors applications created with the use of classical development tools or scientific computing environments of Matlab type. Most often, in these papers, there is a lack of studies which use the synthetic models with a priori known parameters, structure and other characteristic spatial features which confirm the effectiveness and verification of used algorithms. Therefore, the article provides an overview of segmentation methods and analysis of possible data filters and results of the development include their use in relation to the prepared and strictly (exactly) parameterized synthetic model which simulates the points cloud from LiDAR. The Laplacian of Gaussian, Sobel, Canny and Prewitt operators and Robert's cross edge detection are subject to the detailed analyses as well as individual stages of data processing into the raster nature. There are also possibilities of chain use of Freeman code and Douglas-Peucker algorithm for the vectorization of the obtained data. The authors present an assessment of the effectiveness and registered problems. The works were summarized with the preparation of the author's application aimed to the segmentation and the creation of the digital terrain model. This application was implemented to work on real data (empirical analyses).