Image Dehazing Method Based On EPI Light Field Depth Estimation And Atmospheric Scattering Model

Image dehazing has always been one of the research hotspots in the field of computer vision.With the advancement and development of science and technology,the requirements for clear images are becoming more important.How to recover clearly visible haze-free images from haze images has become the focus and difficulty of research.Through the atmospheric scattering model,a clear image without haze can be directly recovered from the haze image,and through the in-depth study of the atmospheric scattering model,the role and importance of the image scene depth information for image dehazing are found.In a real haze scene,the scattering effect of haze droplets in the air on the light,and considering the attenuation of the light itself during the propagation process,at the same time,the existing methods are difficult to directly solve the scene depth information of the haze image.The lack of guidance of the image scene depth information during the haze process will cause problems such as color distortion and halo effects in the restored haze-free image.Therefore,this article mainly discusses and studies such problems.The main research work of the thesis is as follows:(1)Explain the development and background of image dehazing,as well as the current research status at home and abroad,analyze the research significance of this topic,and summarize and summarize the existing image dehazing methods according to different principles.The principle of haze image imaging and the theory of dark channel prior dehazing are introduced in detail.At the same time,the problems and solutions of dark channel prior are explored.(2)Aiming at the problems of color distortion and halo effect in image dehazing,a transmission calculation method combining light field depth estimation and dark channel prior is designed.The scene depth information is obtained through light field depth calculation,and then using the scene depth information to directly calculate the fusion of the initial transmission image and the dark channel transmission image to obtain the final transmission image,which can effectively improve the quality of the transmission image,and through experiments compared several image dehazing methods,verified the depth of field information in the image With the participation,it can have a good effect on the color restoration of the image after dehazing and the suppression of the halo effect.(3)Aiming at the problem of atmospheric light calculation in image dehazing,a sky area segmentation algorithm based on scene depth is designed.Using the scene depth information obtained by the light field depth calculation as a guide,the sky area of the haze image is segmented,and the atmospheric light is calculated separately for the sky area.Color correction is performed on the defogged image according to the information in the sky area,and the effectiveness and accuracy of the method in this paper are verified through comparative experiments.