Vehicle's Model classification using a Vertical Stereo Camera

摘要

This paper proposes a method of vehicle's model classification based on stereo vision. First, we find features that horizontally separated shape of vehicle are selected because of coming into view well after matching. After extracting a side outline of vehicle using a vertical stereo camera, vehicle is classified with normalized vehicle's front-side image. As a result of experiment, our method shows an improved 1.62 [%] compared with a method using a vehicle's front-side image only.