Research And Implementation Of Public Bicycle Parking System Based On Machine Vision

At present,in order to reduce the cost of parking site,the public bicycle parking system is changing from the traditional "one car one pile type" to "no pile type".However,without the limit of bicycle pile,the situation that people park public bicycles casually can happen everywhere.So it's very urgent to build an advanced parking management system relying on information technology.According to the current problems in the bicycle parking system,the machine vision technology can be used to carry out real-time,automatic and all-weather supervision in the system.Firstly,in order to solve the problem of vehicle identification and tracking and positioning in parking system,this paper deeply studies the existing image registration and moving target tracking algorithm,and puts forward the improved algorithm.At the same time,based on the research results,this paper designed the no pile public bicycle parking system.The test results show that the parking system can effectively identify and track the bicycle entering the parking station,and realize the intelligent management of the public bicycle.The main contents in this paper are summarized as follows:(1)In order to achieve the rapid identification of the vehicle,this paper study about the SURF(Speeded-Up Robust Features)algorithm.Aiming at the problem of traditional SURF algorithm,such as slow speed of recognition and the error matching,this paper proposed an improved SURF algorithm combined with the image entropy.This new method improved the efficiency of the algorithm effectively.At the same time,this paper introduced a two-way matching mechanism which reducing the amounts false matching point pairs effectively;it improved accuracy of the registration of public bicycle.(2)In order to tracking and position the public bicycles in the parking station,this paper study about Camshift(Continuously Adaptive Mean-SHIFT)algorithm.Aiming at the problem that Camshift algorithm often tracks false object in complex background easily,this paper proposed an optimized Camshift algorithm.In the first place,algorithm detects moving targets through the background subtraction,at the same time,removing the interference pixels in the tracking frame and ensure the tracking accuracy according to Pakistani distance images before and after frame histogram.Experimental results show that the algorithm can track the target effectively and improved the location accuracy.(3)Accroding to the research results above-mentioned,this paper designed a public bicycle parking system based on machine vision.This system contains five modules includes the main board module,the GPRS module,the card reader module,the memory module and the camera module.The image matching and target tracking algorithm proposed in this paper are tested on the hardware system,and the paper analysed the experimental data.The experimental results shows that the public bicycle system designed based on machine vision can identify and track the vehicle effectively.It realized the intelligent management for bicycles successfully.