Implementation of Smart Parking System Using Image Processing

Today, most of Sri Lanka’s operational parking areas are controlled manually and inefficient. For this reason, drivers can take extra time to drive around the parking area in busy days to find a free parking spot. The proposed system can identify the vehicle type and free parking slots. Also, the system contains the shortest path algorithm to help drivers in finding the nearest vacant parking space. In this project, the camera is acting as a sensor. The reason behind using a camera is with an image it can detect the presence of many vehicles at once. Depending on the area to be covered, one or more cameras are used to process video clips. Since there are no sensors employed, the mechanical and electronic functionality of the system is reduced to a great extent. In previous research, Canny Edge detection method was used to identify the vehicles. In this research, Yolo algorithm is used to detect the presence and the type of vehicles in the parking area because Yolo algorithm is more accurate compared to Canny Edge detection method. After detecting the vehicles, the mean pixel value of each vehicle is taken. The coordinates of the parking slots are taken to an XML file when the parking area is empty and mark the polygon area of each parking slot. Ultimately, each parking slot is uniquely numbered, and status of the parking slot is decided according to the mean value of the vehicle. If the mean pixel value of the vehicle is inside the given bounding box then, that parking slot is not empty. Otherwise, it is empty. The updated status of each parking slot is displayed at the entrance of the parking area. The accuracy of system was approximately 95% in different weather conditions and various angles of images. The smart parking system reduces the stress, waste of time associated with the parking vehicles and makes managing such parking areas more economical.