Nodal points in the Central Area of the Warsaw University of Technology were analyzed using specific cameras with built-in image analysis systems as well as currently installed monitoring (CCTVs). The first mentioned apparatus was dedicated to extracting and tracking pedestrians and vehicles. The second, collecting images which were also analyzed. This allowed them to distinguish and detect different objects and analyze their motion. The main limitation was, a significant number of cameras (with high unit cost), required to be installed to cover all subjected area which results from both restrictions on the installation spots for cameras as well as significant forest cover.
Fig. 1 Preliminary results of measurements of the presence of pedestrians in the PW car park, using cameras with built-in image analysis (yellow dots)
Fig. 2 Preliminary results of measurements of the presence of cars in the PW car park, using cameras with built-in image analysis (yellow dots)
Fig. 3 View of the simulation of camera placement, taking into account simplified tree models in the analyzed area.
As an alternative solution, the images from the monitoring cameras already installed in the analyzed area were tested. This gave satisfactory final results, with a significant reduction in the cost of installing the system. One of the key challenges in this method was to reduce the time and required computing power, while maintaining acceptable accuracy of data acquisition. In addition, the challenge was to accurately determine the point of contact between the pedestrian and the ground, for the purpose of further analysis of data on campus maps, in the GIS space. These issues are presented in more detail in the tabs: Data processing and analysis and Data transfer in a GIS environment.
Fig. 4 Examples of analyzes of the presence of pedestrians obtained on the basis of recordings from surveillance cameras.