Research On The Traffic Sign Recognition System Suitable For Hazing Weather

In today’s society,the Traffic congestion problem has severely restricted the development of the city.Most of the scholars put forward using intelligent transportation system to solve this difficult problem.Traffic sign recognition system is an indispensable part of the intelligent transportation system.But under the haze weather,the environment of traffic image degradation is seriously weakened,which cannot guarantee the accuracy of recognition system.On the other hand,Mobile devices are indispensable in our daily life.Among which the mainstream of the market is using Android operating system.Therefore,within this article,by study on physical model of degraded image in the fog haze weather,the single traffic image fogless algorithm is raised based on the sky region segmentation.And this article explored the traffic sign detection and recognition method,verified the validity of the algorithm.Eventually the system is transplanted into the Android platform,which could be a preliminary implementation that apply the haze weather traffic sign recognition system on mobile devices.Firstly,there is a fact that the existed algorithm is easy to cause partial color in transition region when it is directly applied in the haze large sky area.It cannot meet the requirement of the subsequent application of traffic sign detection system.In this article,based on the principle of the classic dark grey prior,it proposed to improve the method of traffic image to haze by using the improved mean shift algorithm for traffic image segmentation.And then the algorithm of histogram equalization to haze in the sky area is applied to remove the fog.The non-sky part is processed based on fast the dark color of the prior bilateral filtering algorithm.Finally,clear image is obtained by image fusion.It is verified by simulation experiments that compared to other several typical algorithms,the proposed algorithm can get an improved image for traffic signs in the transition area between the sky area and color distortion.And by quantitative analysis the recovery effect is better.Secondly this article identified traffic signs by color-shape detection to build the interested area.The HOG features of samples are dimensionality reduced by use of PCA technology.The SVM is obtained by training to classify traffic signs.This paper establishes the traffic image database for degraded images.Through designing thedifferent detection process,detect and analysis effect that the above de-fog algorithm is used to the traffic sign detection,The algorithm using experimental results in this paper shows that the recognition system for traffic sign recognition rate is higher than other identification systems that with classic fogless algorithm in the database.Lastly the test identification system is transplanted to the Android platform.It primarily achieved traffic sign detection recognition system application adapted to haze weather.