Outdoor vacant parking space detector for improving mobility in smart cities

Abstract

Difficulty faced by drivers in finding a parking space in either car parks or in the street is one of the common problems shared by all the big cities, most of the times leading to traffic congestion and driver frustration. Exploiting the capabilities that Computer Vision offers, an alternative to those ITS commercial solutions for parking space detection that rely on other sensors different from cameras is presented. The system is able to detect vacant spaces and classify them by the type of vehicle that could park in that area. First of all, an approximate inverse perspective transformation is applied for 2D to 3D reconstruction of parking area. In addition, feature analysis based on Pyramid Histogram of Oriented Gradients (PHOG) is carried out on every parking zone within the parking area. Experiments on real scenarios show the excellent capabilities of the proposed system with independence of camera orientation in the context. © 2013 Springer-Verlag.