Recognition of Cross Section and Vehicle Classification from Single Image

Abstract

In this paper using an image shot by single camera located at the side of road, we proposed a method to recognize the shape of a vehicle and classify the type of that vehicle.
Primarily it is impossible to reconstruct the shape of a vehicle from single image. Therefore proposed method makes it come true with information that the shape of the rear has kind of symmetrical. And the method classifies vehicles using template matching based the tilt of recognizable data of cross section.
Experimental results with simulation images and real images show that proposed system is efficient.