Surround view based parking lot detection and tracking

Abstract

Parking Assistance System (PAS) provides useful help to beginners or less experienced drivers in complicated urban parking scenarios. In recent years, ultrasonic sensor based PAS and rear-view camera based PAS have been proposed from different car manufacturers. However, ultrasonic sensors detection distance is less than 3 meters and results cannot be used to extract further information like obstacle recognition. Rear-view camera based systems cannot provide assistance to the circumstances like parallel parking which need a wider view. In this paper, we proposed a surround view based parking lot detection algorithm. An efficient tracking algorithm was proposed to solve the tracking problem when detected parking slots were falling out of the surround view. Experimental results on simulation and real outdoor environment showed the effectiveness of the proposed algorithm.