Research On Fog Removal Algorithm Based On Dark-light Channel Prior Fusion And Adaptive Parameter Optimization

Digital image processing system is involved in every field of our daily life.Due to the influence of environmental factors in recent years,frequent haze weather leads to the serious deterioration of image quality,which brings great challenges to the subsequent image processing technology.Therefore,it is of great significance to input foggy image for fog removal processing and output high quality foggy image.Although the current mainstream dark channel prior-based defogging algorithm shows an ideal effect,it also has some defects,such as inapplicability to the sky area,inaccurate estimation of atmospheric light value and transmittance,complex refinement process of transmittance that leads to a long calculation time and inadaptability of defogging parameters.Based on the National Key Research and Development Program of China(No.2019YFE0108300)and the National Natural Science Foundation of China(No.61302150),In this thesis,a new defogging algorithm is proposed to improve these defects and achieve good defogging effect.The main innovation points of this thesis are as follows:1.In order to improve the estimation accuracy of atmospheric light value and transmittance,a new atmospheric light scattering model is proposed in this thesis,which combines dark channel prior theory with bright channel prior theory to obtain more accurate atmospheric light value and transmittance.2.In view of the problem that the refinement process of transmittance is complex and leads to a long calculation time,when refining the initial transmittance,the gray image of foggy image is taken as the guiding image for guiding filtering in this thesis,which can greatly reduce the time complexity of the algorithm.3.In order to solve the problem that dark channel prior theory fails in the sky area,the method of using compensation function to improve the transmittance of the sky area adaptively,which improve the adaptability of the algorithm and effectively solve this problem.4.Due to the lack of adaptability to the parameters in the process of defogging,a method to calculate the intensity of fog removal by atmospheric light value is put forward in this thesis through lots of experiment,meanwhile,this thesis presents an improved algorithm of adaptive filtering window,which can calculate the optimal minimum filtering window and guide filtering window,obtain the best restored image.In order to verify the effectiveness and reliability of the defogging algorithm proposed in this thesis,some classic defogging algorithms as well as the defogging algorithm with excellent defogging effect in recent years are selected to carry out comparative experiments with the proposed algorithm.The experimental results show that the proposed algorithm is superior to other algorithms in both subjective and objective quality evaluation.