Research Of Intelligent Parking Management System Based On Multi-camera Coordination

With the social development and the improvement of living standards, our country has a rapid growth in the private car ownership. It results in traffic congestion and a substantial increase in parking demand, so the requirements of extending the scale and meliorating the environment of the parking place are increasing rapidly. After the completion of the parking lot layout, it becomes more import for facilitating the parking of vehicles, improving the vehicle safety, reducing the operating cost and improving the management of the car park. Nowadays, Most of the current management system can only manage the entrance for the car park effectively, but can not obtain the inside information easily. It also can only provide the time charges, and can not support the large numbers of the cars and the large area of the car park. It is unable to display the global dynamic visual information of the car park, so it is a great gap between the actual demand and the reality.This article describes an overview of the car park management, and the structure and development of the system. We describe the key technologies and discuss their respective advantages and disadvantages, and a new Intelligent parking management system is proposed which bases on a collaboration of the multi-camera mode.1,According to the conditions of the distribution of parking spaces and the requirements for intelligent parking, we design the intelligent parking system and introduce the system's hardware configuration and software modules. Through the dynamic updates of the parking condition, we can manage the vehicles in the car park effectively.2,We propose the method of detecting the dynamic state of the parking place, and import noise by dealing with the video images firstly. The second, we calculate the location of the maximum number of occurrences of pixels as background pixels, then build the initial background model of the parking lot within the range of parking spaces which users select manually. At last, we focus on the pixel value within the scope of the changed circumstances, and use the inter-frame difference method to determine the status of the parking lot and whether the background updates by a combination of these two value changes. 3,We propose the strategy of tracking by collaborative multi-camera mode, which is based on the relationship between the camera and the vehicle. In a multi-camera case, as each camera is fixed, so that their monitoring range is also certain. We attain the projection matrix of each camera by marking the camera by sequence, and record the vehicle status information within a single camera; if the vehicle overlaps in a multi-camera visual domain, we derive the vehicle status information according to the smallest error of the method; if the vehicle is out of range, we estimate the vehicle state information according to the Kalman filtering prediction.We take multi-camera collaborative strategy for monitoring the overall situation of the parking lot and solve the parking management system in a large parking lot which is in the existence of the safety and the problem of low efficiency with simple installation, low cost, real-time monitoring and user-friendly interface.