System for Detecting Vehicle Features from Low Quality Data

Marcin Dominik Bugdol
Faculty of Biomedical Engineering, Silesian University of Technology, Roosevelta 40, 41-800 Zabrze, Poland

Pawel Badura
Faculty of Biomedical Engineering, Silesian University of Technology, Roosevelta 40, 41-800 Zabrze, Poland

Jan Juszczyk
Faculty of Biomedical Engineering, Silesian University of Technology, Roosevelta 40, 41-800 Zabrze, Poland

Wojciech Wieclawek
Faculty of Biomedical Engineering, Silesian University of Technology, Roosevelta 40, 41-800 Zabrze, Poland

Maria Janina Bienkowska
Faculty of Biomedical Engineering, Silesian University of Technology, Roosevelta 40, 41-800 Zabrze, Poland

Abstract

The paper presents a system that recognizes the make, colour and type of the vehicle. The classification has been performed using low quality data from real-traffic measurement devices. For detecting vehicles’ specific features three methods have been developed. They employ several image and signal recognition techniques, e.g. Mamdani Fuzzy Inference System for colour recognition or Scale Invariant Features Transform for make identification. The obtained results are very promising, especially because only on-site equipment, not dedicated for such application, has been employed. In case of car type, the proposed system has better performance than commonly used inductive loops. Extensive information about the vehicle can be used in many fields of Intelligent Transport Systems, especially for traffic supervision.