Research And Design Of Enhanced Parking Space Identification System

With the development of technology of computer, image and communications,the method of solving problems related to the transport area by using scientific andintelligent has begun to attract more and more attention from researchers in the fieldof transportation. In this background, ITS (Intelligent Transport System) to solve avariety of problems of the transportation area become an ideal choice.A lot ofcountries attach importance to the development and construction of IntelligentTransportation System, and in some regions has been got rapid development andapplication.Intelligent detection of parking spaces attract more and more attention as part ofintelligent transportation systems,especially parking space detection technology basedon the theory of image recognition provides a new solution for intelligenttransportation system. Compared to other technologies, image detection technologyhas unparalleled advantages, camera is installed convenient, the position of thedetector is easily adjusted and moved and so on.In order to solve the problem of parking space management,this paper study the parking detection system,which combines the current technology of embedded,database and image recognition and provides a set of information technology to achieve parking management reference scenario for parking detection system.The system in this thesis is based on the top-down science method of embeddedsystem.Firstly the parking space detection algorithm is analyzed, Secondly therequirement of parking detection system platform and system is analyzed,thirdly themain function of the system is designed. Finally, the optimization algorithm is carriedout based on the hardware platform, the calculation speed is improved.The mainresearch contents are as follows:(1) According to the practical,parking space detection algorithm is integrated andoptimized, the underground parking characteristics, using new corner features, at thesame time in the system of SVM is used as a recognition method.(2) System requirements based on embedded Linux system kernel has been cutand transplant.(3) The host-side applications is designed and developed Visual Studio2010,the terminal software is designed and database is designed using MySQL.(4)The image processing algorithm is transplanted in the Cortex-A8platform,theimage algorithm set is optimized by used NEON instruction and tested by contrast.