| Abstract [eng] |
The aim of the Master's thesis is to isolate all parameterization algorithms and their influence on vehicle classification algorithms. use the reflected signals measured by the millimeter-wave radar as the vehicle passes by the field of vision of the millimeter-wave radar. The first part of the work analyzes the literature. Non-contact sensors for non-invasive vehicle performance evaluation are reviewed. More vehicle classification standards are compared for their advantages and disadvantages. The second part describes the data collection system, its parameters and measurement methodology. Describe how the frequency modulated signal used during operation was calculated and the importance of each parameter. The third part describes and analyzes the specialized DBScan algorithms used to parametrize vehicles, i.e. to obtain vehicle parameters. Their performance, purpose of use, advantages, disadvantages are analyzed. In the fourth part, a preliminary data analysis is performed to remove the incorrectly calculated coordinates from the reflected part of the vehicle. The fifth examines the parameters of the DBScan algorithms. In the sixth part, specialized DBScan algorithms are investigated for their influence on the clusterized and subtracted parameters (length, width, reflected signal strength, number of NTKPAS, velocity) of the reflected coordinates of the reflected signal forming the TP. In the seventh part, vehicles are classified using all parameters obtained by clustering algorithms. The importance of the parameterization algorithm and vehicle classification is investigated. |
