Detecção e estimativa de velocidade veicular através de processamento de imagem

Abstract

Image processing techniques are increasingly present in vehicle traffic monitoring. This final paper proposes an initial design of a vehicle speed meter using only a fixed camera and a computer. Firstly, it was introduced the images forming process and images properties that wil be relevant to this final papers. For the algorithm criation, it was used diverses images process techniques, such as background subtraction, morphological operation, contrast equalization and perspective transformation with the objective to increase the vehicle tracking chance and the speed calculation. The obtained results was compared with reference values of inductive traditional meters in order to validate the method. The vehicle detection rate is around 95 % and the average speed error is ± 2,265 km/h.