Research And System Implemntation Of Dehazing Method Based On Color Attenuation Prior

In the bad weather conditions,such as fog and haze,the visibility of outdoor scenes drops sharply,resulting in the serious degradation of images or videos captured by Imaging equipment,such as faded color,loss of details,and reduced clarity,which affects and limits the normal operation of the video surveillance system.Therefore,how to improve the dehazing effect of images and videos in a video surveillance system has very important practical significance.Based on the research of single image dehazing algorithm,this thesis proposes a video dehazing method suitable for outdoor surveillance systems,and designs and implements two different types of dehazing systems.The main research contents are as follows:(1)An improved color attenuation prior dehazing algorithm is proposed.Firstly,the nonlinear scene depth model is used to replace the original linear model,and on this basis,the filtering optimization is carried out to recover more accurate scene depth;Secondly,by defining a function of the product of the depth of field and the depth of field index,and using a large number of image experiments to obtain the optimal value of the function parameters,so as to obtain a more accurate atmospheric scattering coefficient;Finally,by increasing the strategy of comparing adjacent areas,the selection area of quadtree division is optimized,thereby improving the accuracy of atmospheric optical value selection.(2)A video dehazing method based on universal transmission is proposed.This method first extracts the common background image in the video according to the characteristics of the relatively static background of the monitored video by using statistical average background modeling method;Then,the improved single image dehazing method is used to obtain the atmospheric optical value and transmittance map of the common background image;Finally,the atmospheric optical value and the transmittance map are used as the general atmospheric optical value and the general transmittance map of each video frame,and are applied in the dehazing process of each video frame,so as to realize the dehazing of the surveillance video.The experimental data results show that the method in this thesis can effectively improve the processing effect of video dehazing,and can basically meet the real-time processing requirements of video dehazing.(3)On the basis of the above image and video dehazing methods,a single image dehazing system based on Matlab R2014a development platform and a video surveillance dehazing system based on Visual Studio 2013 and OpenCV 2.4.9 development environment are designed and implemented.Both systems are integrated with the dehazing method and some classic dehazing algorithms in this thesis.