Expert system design for vacant parking space location using automatic learning and artificial vision

Abstract

Finding a free parking space nowadays is a recurring problem in increasingly crowded public parking lots. The present study offers a solution that is based on the analysis of zenith images using artificial vision and is capable of automatically analyzing both the available spaces in the parking lot and their real-time occupancy. In an initial phase, the presented system semi-automatically detects the available parking spaces by filtering, thresholding, and carrying out a process of extracting the contour and approximating to a polygon the parking spaces of an empty parking lot. Once the size and location of the parking spaces have been mapped, the system is capable of detecting not only the presence of a vehicle in a parking space, but also the area of the parking space occupied by it with an accuracy of 98.21% using Region-based Convolutional Neural Networks. This feature allows the system to specify the appropriate parking space for a new vehicle entering the parking lot based on its specific dimensions and the correct location of the cars parked in the spaces adjacent to the free space.