Search

Projection of a complicated geometry of industrial objects is the complex issue, which requires properly planned and prepared measurements. Such objects must be accurately inventoried, but their complicated nature often makes the access and the visibility of their entire surface very difficult. Documentation of measurements is often prepared in the form of sketches, plans or maps, which are amended with photographic documentation. The objective of this paper is to test the possibilities to apply laser scanning and the network of digital images for inventory and monitoring of technical conditions of industrial objects. Processing of a precise documentation acquired basing on terrestrial laser scanning data or dense points clouds generated from digital images still causes many difficulties and problems. Although data processing algorithms have been intensively developed with respect to generation of high resolution orthoimages or precise vector drawings, the existing problems are still connected with limitations related to imperfections of both techniques of measurements.

In this research, the issues related to processing and analysis of the images, acquired by the lowcost terrestrial sensors and an unmanned aerial vehicle (UAV) of the archaeological sites in an urban environment placed in area of Kutaisi, Georgia, have been considered. Taking into account generation of the Bagrati site photogrammetric documentation, the low-cost UAV DJI Phantom 3 and Pentax SLR camera, were chosen. Due to the geometrical and radiometric parameters of the chosen sensors, an extended methodology of UAV and terrestrial data processing in Agisoft PhotoScan was proposed. This task solution was enabled by application programming interface API (which allowed taking advantage of software functions - not available in the basic version of the program) and additionally the script written in MatLab software by authors. The method of processing, filtrating, eliminating and weighting of the measurement observations in the bundle adjustment process was proposed. As a result of performed works, the accuracy of the results of the bundle adjustment, increased significantly in comparison to the raw data. Furthermore, statistical data analysis had been expanded. [ABSTRACT FROM AUTHOR]