Research On Road Traffic Sign Recognition System Based On Machine Vision

With the popularity of cars,the urban traffic safety problem is becoming more and more significant.The intelligent transportation is developing in such circumstances.Among all these road traffic information,road traffic signs provide direct instructions to drivers.Therefore,the extraction of road traffic information is particularly critical.In this paper,the fast extraction and accurate identification of road traffic markings are systematically and deeply studied.With the help of image processing,machine vision,pattern recognition,statistics and other relative theories,the Android based intelligent mobile terminal is used as the platform to develop a set of road traffic signs recognition algorithms.It is then integrated as one of the core functions in the driving assistant application of the intelligent mobile phone,which is used to process the images captured by it mounted on the moving vehicle and gives out the necessary warning for drivers.The object of this study mainly including the orientation marks and the lane marks.The main results of this study are listed as follows:(1)This study proposes a method for determining a region of interest: Extract the information from the complex background crudely.According to the location and characteristics of the research object,the region of interest is determined,which can not only reduce the data redundancy,improve the processing speed of the system,but also reduce the noise interference and improve the recognition accuracy.(2)Calculate the gray histogram of the image,introduce the idea of centroid,and analysis the statistical properties of the histogram,a new binarization algorithm based on the centroid of the histogram is proposed,which could adjust the threshold adaptively(3)In this paper,an improved template matching algorithm is proposed,which is used to identify the irregular shape with affine deformation and convex concave edges.According to the color feature and shape feature of the target,the algorithm is improved from three aspects in order to achieve fast and accurate identification of road traffic signs,which are template library construction,feature extraction and similarity measurement.(4)Design the overall scheme and technical route of real-time detection of road traffic marking,and use it as one of the core functions to be applied to the intelligent mobile assistant driving system based on Android system.A large number of real tests were carried out in different natural environments,different driving speeds and different road sections.The success rate was 90% and the average processing speed was 90ms/frame.The speed,accuracy,robustness and stability of the system can meet the needs of usersThe results of this study not only provide a new method for the identification of irregular shape,but also integrate it into a function of intelligent mobile terminal applications which was realized and tested through a large number of driving on the real road.The experimental results verify the feasibility of the algorithm,it opens up a new way for the application of machine vision theory in intelligent transportation system,and it is a meaningful attempt in the intelligent mobile terminal technology development in the field of the driving assistance system.