| The countries all over the world have always focused on the study of intelligent transportation system. They have made great progress in easing traffic congestion, enhancing driving safety and managing vehicle intelligently. Parking space management, as an important branch of intelligent transportation system, has the functions of license plate recognition and memory, vehicle tracking, vehicle parking service, parking charge by and so on. Based on the parking management system, studying on the states of parking spaces, we design a parking space detecting system.First, this paper presents a parking space detection algorithm. Different from sensors, the algorithm considers parking video images as the research object. The parking images are obtained by the camera fixed at the parking. The images will get pretreatment firstly, pretreatment including the removal of environmental noise and the conversion from RGB color space to HSI color space. Then it selects the parking space detection area. Based on the pixel color sample mean and covariance matrix in advance, it uses Gauss function to classify the pixels in the testing area, and calculates the transmission probability of adjacent parking state. As the two kinds of state for nodes, and as the reciprocal of transmission probability for branch weight, it builds a parking spaces status combinatory model. Based on MAP, it finds a set of best combination, and more, obains the parking status information.The algorithm is transplanted to embedded Linux development platform. At first, we establish the cross-compiler environment in the Linux PC system. We debug and compile the program in PC, and with the help of NFS, download it to Rea16410. Based on C/S network model, we can transfer the detection result to the server. Users can get the result through interactions with the server. Using the method of I/O multiplexing technology, the paper solves the data concurrency issues when many users accessing the server at the same time.Experiments show that the system can eliminate the effects caused by occlusions of neighboring pairs well. And it can effectively handle the data concurrency when some users accessing the server at the same time. |
