The Speed Limit Sign Recognition Based On Machine Vision Technology Research

With the development of social economy and the continuous improvement of people's living standard,the car ownership keeps increasing,and the increase of the car penetration rate also makes the road traffic bearing more and more pressure.In order to alleviate traffic pressure and reduce the loss caused by traffic accidents,the development of intelligent driving assistance system which can extract key traffic information in real time and make early warning has become a research hotspot.This thesis chooses the automatic detection and recognition of speed limit markers as the research purpose.Comprehensively use theoretical knowledge such as machine vision,image processing,pattern recognition,statistics to process images acquired by vehicles traveling in real scenes.The main work and outstanding contributions are as follows:(1)A speed limit sign detection method based on improved Sobel algorithm is proposed.The edge binary image is obtained by this algorithm,and the edge binary image is segmented to obtain multiple connected regions.Based on the geometric characteristics of the speed limit sign,the connected areas are screened and the suspected target area is determined.(2)A similarity measurement algorithm based on projection features and a corresponding speed-limiting sign recognition technology are proposed.The projection feature is obtained by bidirectional projection of the target number,and then the similarity coefficient between the target number and the template feature is calculated by the signal similarity measurement algorithm.Classification and recognition of digits are realized based on similarity coefficient.(3)Based on the existing Android smart mobile terminal platform,real-time vehicle testing is carried out.More than 300,000 pictures were tested in different natural environments and road conditions.The average running speed is 150 ms/frame,which meets the real-time requirements.The success rate of comprehensive detection of various environments is more than 95%.Experiments show that the method proposed in this paper has high recognition rate and can be well adapted to speed limit sign recognition under various external conditions.It provides a feasible path for the application of auxiliary driving system in intelligent mobile terminal,and provides a reliable theoretical basis and experimental data for further research on intelligent extraction of traffic information.