METHODOLOGY FOR ASSESSING THE QUALITY OF POINT CLOUDS BASED ON DENSITY CONDITIONS

This article discusses a numerical method for estimating the dependence of the final quality, such as the density of a point cloud per square metre of pavement, on the speed of movement of a mobile laser scan. In practice, long roads are most often scanned using mobile laser scanning technologies (or hybrid with photogrammetric data), whose outputs are dense point clouds with the need for classification, panoramic images and other ancillary data. As part of the analysis, Voronoi diagrams are used to divide the target road contour into separate sections for subsequent calculation of how many points fall within each of the polygons. For the analysis we developed an original methodology using Python automation tools and QGIS, PostgreSQL + PostGIS tools, based on the idea of checking the average point density for the target road surface divided into analysed sections (polygons Voronoi). The materials for the analysis are initial point clouds, stitched and filtered for several urban roads of St. Petersburg with a total length of 3.1 km, and a set of polygons in the form of Voronoi diagrams. When analysing the data, it was found that the range of point densities 0...1000 pcs/sq.m. contains the largest group of polygons in terms of area. Numerically, it represents 30 % of the scan volume. According to the results of the study, when creating a digital model of traffic organisation, the density of 277 points per square metre (which corresponds to approximately 1 point per 6 square cm.) is necessary for the precise identification of technical means of traffic organisation. This accuracy can be achieved when the scanning vehicle is travelling at a speed of up to 35 km/h, provided that other surveying rules are observed. However, a lower density is possible for tasks other than checki the road for rutting or determining the exact pattern of horizontal road markings.