Research Of Recognition Method Of Speed Limit Sign In Driving Behavior Evaluation System

The research of driving behavior is the core problem in road safety. The research-ers analyze the collected driving information and build models to predict driving be-havior in order to reduce traffic accidents and casualties and economic losses. With the development of computer technology, large-scale data acquisition has become possible. However, it is a difficult problem that selecting and analyzing specific data required from the massive data. It will effectively enhance the efficiency of screening data by using computer technology to analyze collected data automatically.Speeding is an important part in driving behavior and a speed limit sign recogni-tion method based on the existing theoretical basis is proposed in this paper. Firstly, the region in image which needs to be processed is determined by regional search strategy and then the SVF (Simple Vector Filter) is used to filter the image to remove unwanted achromatic color information. Secondly, the color segmentation of image using multi-channel and the binarization of image is implemented and the traffic signs which may be adhered are divided. Then the skeletons of image are extracted and detected by CMHT (Chord Midpoint Hough Transform) method. Thirdly, the traffic signs which are not speed limit signs are excluded by the statistical characteristics of traffic signs in or-der to determine where the regions of interest (ROI) are. Finally, the numbers are ex-tracted from ROI and the tilt correction, character segmentation, normalization and skeleton extraction are implemented. Then the digits are recognised using the handwrit-ten digit recognition by projection in seven segments frame "8" and the output is used for the driving behavior evaluation system.An interactive driving behavior evaluation system based on database which in-cludes multiple functions is developed according to data collected by researcher and the requirements of project and the speed limit sign recognition method above is applied in the system. The information of driving behavior which is collected by devices is stored in database and displayed on an integrated interface which is similar with real driving conditions by driving behavior evaluation system, allowing researchers to quickly navi- gate to the required driving video segments to browse and evaluate the information of driving behavior in database.