Research On The Dark Channel Fog Removal Algorithm For Road Monitoring

With the rapid development of science and technology,computer vision are playing a more and more important role in people's daily life.In the application of monitoring,some factors such as low visibility,haze weather will lead to the degraded image and further influence the retrieval of the original vehicle information,which greatly affects the computer vision system processing image information.So in the background of an increasingly deteriorating environment,frequent presence fog and haze,the significance of image defogging technology is self-evident.In this paper,image defogging of the road monitoring images is studied.First,mainstream image defogging algorithms are analyzed.Then,an adaptive image defogging algorithm is proposed and experimentally studied.The main contents of this paper include:(1)The effects of haze on the image of the object is analyzed based on the atmospheric scattering model,which explains why images are degraded in haze weather.An SVM classifier is proposed to judge if an image is a haze image.This solves the problem of manual intervention before image processing.It can be seen from the experimental results that this algorithm can be automated to distinguish haze images and good results are achieved with respect to time and accuracy.(2)The image defogging algorithm based on physical model are analyzed.The algorithm for single image defoggings based on the theoretical basis are mainly studied and,main problems existing in this algorithm are analyzed from the view of dark channel.Some model based defogging algorithms are compared with each other through experimental results and the operation speed.(3)To solve the problems such as time-consuming,distortion,dark restored image of the dark channel prior algorithms,this paper proposes an improved algorithm to fog the dark color: the original haze image by wavelet transform method using adaptive block for dark channel image and low frequency components.Get the rough distribution of haze image transmission after the inverse wavelet transform;then fast bilateral filter was used to smooth the rough transmittance image;the dark channel transmission of the failure zone is modified to solve the distortion problem;finally the contrast enhancement algorithm is used to increase image brightness without changing the image tone.Experimental results show that the improved algorithm of fast dark first prior fog is more effective than the original one with respect to time consumption and defogging effects.Experiments of the proposed algorithm and the original algorithm are carried out in this paper with the help of MATLAB.Experimental results show that the proposed algorithm has better performance in both subjective and objective metrics.The algorithm proposed can be used as a preprocessing algorithm for monitoring road vehicle information retrieval,identification and vehicle driving recorder and so on.In the real-time monitoring of city road condition,traffic safety contributed to haze and other adverse weather conditions will be ensured and the incidence of traffic accidents will be reduced.