The usage of geospatial tools in traffic sign detection

Abstract

The topic of this scientific paper is the detection of traffic signs based on the data obtained using modern geodetic methods (LiDAR, mobile laser scanning), by using geospatial tools. In a total of 100 photos taken in an urban area in Germany, in a length of 500 m, 75 traffic signs were detected. Each sign is assigned a corresponding code from the catalog of German traffic signs, after what the photos are connected to the point cloud. Then the extraction of bounding boxes and central points of traffic signs was done. The visualization was done in two softwares -Bentley MicroStation and QGIS, which completed the detection process. The accuracy of traffic sign detection in photos is 83% (62/75). The extraction of bounding boxes and central points on the point cloud was performed with a high degree of accuracy (85%).

Keywords

• traffic sign detection
• geospatial tools
• geodesy
• geoinformatics