Vehicle recognition based on spatio-temporal image analysis

Abstract

It is important to assess street-parked vehicles causing traffic problems in urban areas. However, it is performed manually and at high cost. It is a top priority for reducing cost, to develop a detection system of those vehicles. We propose a detection method using a line-scan camera mounted on probe vehicles. The method is based on two kinds of epipolar-plane image analysis; one is edge-based algorithm and the other is region-based algorithm. Both algorithms are fused to achieve higher reliability. As a result of experiments in roads, a detection rate reached 96%.