Parking Space Detection Based on Information from Images and Magnetic Sensors

Abstract Parking space detection is a key technology in the parking lot management system. When performing parking space detection using magnetic sensors, there may be interference from the adjacent parking space, which will result in false detection. However there still will be false detection when only cameras are used to detect the parking spaces because the performance of the cameras is inclined to be affected by the environment. Considering that the magnetic sensors are only sensitive to ferromagnetic objects and insensitive to the environment effect and images often contain a wealth of information, we propose a parking space detection algorithm combining the information from magnetic sensors and images. The algorithm performs primitive detection using magnetic sensors. If the sensor reports that there is a vehicle nearby, the algorithm analyzes the image information of the parking space gathered by the camera, extracts the image features of the parking space and classifies the features to confirm whether the parking space is occupied. The experimental results show that our algorithm can make more precise decisions for parking space detection with much lower complexity and less computational efforts. Furthermore our algorithm can effectively reduce the burden of image or video processing of the system.