Vision-based Vehicle Detection and Classification

The competitiveness, economic strength and productivity of a country heavily depend on the performance of its transportation systems since they are an indispensable part of human activities. Road networks are a basic element... [ view full abstract ]

The competitiveness, economic strength and productivity of a country heavily depend on the performance of its transportation systems since they are an indispensable part of human activities. Road networks are a basic element of transportation systems, with an average of 40% of the population spending at least 1 hour on the road each day. Unfortunately, though, the increased vehicle traffic causes a plethora of problems such as traffic accidents, traffic congestion, air pollution and pavements deterioration. The option of constructing new infrastructure and/or the enhancement of existing infrastructure by widening the roads is most often neither affordable nor feasible due to limited land resources. Consequently, the need for enhanced surveillance and for management of traffic was generated. Vehicle counting is already being conducted by transportation authorities either manually or automatically by the use of sensors. The former is tedious and expensive, while the latter may be limited by technology shifts and data compatibility issues, disruptive installations, sensor placement optimization issues and skilled usage. In spite of the existence of a plethora of vision-based algorithms, which accurately detect and track vehicles, the task of classification faces the challenge of the range of shapes and sizes within classes of vehicles. The proposed methodology builds on computer vision algorithms in Matlab and newly developed custom algorithms to detect, track and count vehicles, while it classifies them based on the blob parameters of area, perimeter and diameter. The system utilize Gaussian Mixture Models and low-cost technologies. Videos were acquired by a stationary camera located in roads of the real-life urban network of Nicosia, Cyprus. The accuracy of the algorithm is promising, providing an inexpensive and efficient solution for vehicle counting and classification to transportation authorities.