Research On The Algorithm Of Defogging Based On The Dark Channel Prior

The environmental problems are serious increasingly,the haze weather occurs frequently,it not only threats to the human environment,but also affects the social development,in the fog weather,the image we obtained using devices has a significant degradation,including color distortion,low contrast,unclear details.it seriously affects the subsequent processing including information acquisition,target recognition and other jobs.Therefore,it is necessary to study image defogging techniques.This article mainly talks about the following aspects:(1)In this paper,domestic and international image fog removal techniques have been extensively investigated.Fog removal methods are divided into two categories: enhancement and recovery based fog removal.Enhancement based fog removal methods include histogram equalization,wavelet domain processing,On the other hand,the fog removal algorithm based on the restoration includes partial differential equations,depth of field estimation and prior information.In this paper,a large number of algorithms have been simulated and analyzed detailedly.It is found that the fog removal algorithm based on dark channels and priori knowledge works well in most scenes.Therefore,this algorithm is selected for follow-up study.(2)Some basic theories are studied,including the definition of fog,the conditions of fog formation,the fog imaging model,the fog image characteristics,the dark channel prior principle,the fog evaluation index(subjective and objective).(3)The algorithm of dark channel prior de fog is sudied,and the flow of the algorithm is analyzed in detail,including the calculation of the dark channel,the estimation of the atmospheric light,the estimation of the transmissivity,and the optimization of the transmittance.(4)In the dark channel prior algorithm in the simulation process,found that the deficiencies of each step,and aiming at the problems and defects,for each step of introducing corresponding algorithm and complex in the visual effect and the time of the two aspects of improved.(5)Based on the dark channel prior to the problem of low efficiency of fog algorithm,proposed a fast defogging algorithm based on dark channel prior HIS space,the fog image is converted from RGB space to HIS space and are divided into blocks,for each block for dark channel,and then the dark channel point saturation correction.Then the HIS space was converted into RGB by the mist dark channel component of each point in the image,by interpolation method,global fog component maps,and finally by the fog image minus the global component map,and brightness adjustment,finally get the restored image,The image simulation are implemented on the MATLAB platform,respectively on dark channel prior to fog algorithm and improved algorithm for each step of the simulation,and the fast defogging algorithm dark channel was simulated based on the proposed,through the evaluation index can be seen that the above improved algorithm obtained good results.