Research On Road Signs Supervising And Management System Based On GPS And Image Processing Technologies

With the concept of smart city proposed, more and more research has been done on image processing technology to road signs. However,there is few done to the supervising and management system of highway assets at present. In this thesis, tools such as OpenCV and Matlab has been used to process road signs. At the same time, both the pretreatment standard model and the real-time model for highway signs has been constructed with the point pattern matching algorithm. The standard GPS and images, the real-time GPS and images are associated with the above two models. The possible defects of road signs and their actually locations are obtained by the following expert control algorithm. By the construction of the two models and the application of above processing algorithms, a new kind of supervising and management system to road signs is designed and set up, which is relied on the real-time GPS and image data. This system provides road-sign maintainers with the accurate fixed-point and fixed-timing maintaining logs. It thus can ensure a further intelligent supervision to highway assets.The main research work and achievements of the thesis are as followings:1.The overall design scheme of the supervision and management system to road signs is proposed, based on the real-time GPS and image processing technology. The functional requirements of the system isanalyzed. The main functional structure is concluded and designed. Then,the key technologies such as the acquisition of real-time GPS and images,the data storage optimization and the image pattern matching etc are researched. The pretreatment standard model and the real-time model for highway signs are constructed. The extracting algorithm to best real-time image and the matching algorithm to GPS and images are also designed and implemented. Finally, the simulation and development work for the system are carried out using the tools of OpenCV and Matlab.2.Both the pretreatment standard model and the real-time model have been designed. It prepares for the following image matching process.The pretreatment standard model is established by tagging the standard GPS and images in single sequence. The real-time images and GPS is automatically obtained by highway inspection operations, with the optimized sampling frequency determined. The real-time model is established by tagging the sampled real-time images and GPS in double sequences,.3.The matching method of GPS and images is proposed and the extracting of the best real-time image is realized. The sign matching algorithm is also chosen. The real-time GPSs are tagged in a sequence.The GPS matching threshold is determined on the basis of performance indicators. The best real-time image is extracted through the match searching of the standard and real-time GPS. By extracting thecharacteristics of the standard images, the pretreatment standard images feature database is established. Further combined with the image processing steps of filtering, denosing, tilt correction etc, the matching points matching algorithm is used to compare the real-time and standard images. Finally, the image classification of the real-time road signs is obtained.