Research Of Image Restoration Algorithm Of Night Hazy Image

With the rapid development of image processing technology,it has been widely applied to the fields such as transportation and image acquisition.The acquisition devices are the sources of images acquired.There are many suspended particles in the fog,haze,and rainy environment,which can cause the devices to refract and scatter in the process of acquiring images,and the captured images will have issues of color distortion,reduced contrast,and blurred details.In the application field of image processing,clear and accurate image characteristics are beneficial to the processing results.The image defogging technology has achieved significant results in daytime scenes,but there is relatively little research on image defogging at nighttime.Nighttime images are susceptible to the effects of fog scattering and artificial light sources,which will result in uneven image lighting and severe color cast.The traditional research methods of images at foggy daytimes are no longer suitable for night scenes.Existing research methods often have issues such as serious chromatic aberration,blurred details and obvious halos at the light sources.Therefore,for the shortcomings of existing research methods,this paper improved the image restoration algorithm at night and foggy days.The work is as follows:1.Combining the imaging characteristics of images at foggy nights,a defogging algorithm based on dark channel prior was proposed.First,it built an imaging model for foggy nights,and solved the foggy incident light image and foggy reflected light image of the night scene according to Retinex theory.It obtained the light source position and depth of field based on the image and foggy reflected light images at foggy nights respectively,with the use of the camera imaging mechanism,it found the sum of the distances between the scene point and each light source,and then obtained the fogless incident light images.For the foggy reflected light images at night,the improved dark primary color prior theory was used to solve the fogless reflected light images at nights.It brought the night fogless incident images and night fog-incident images into Retinex theory to solve the night fogless images.Aiming at the issue of color cast in the processed images,the local Shade of Gray algorithm was used to correct the color.The experimental results show that the algorithm in this chapter can achieve defogging well,and the defogging result is clearer,which can effectively eliminate the color cast.2.Aiming at the application of the dark primary color prior algorithm for defogging,the effect is insufficient.In this paper,for the issue of noise amplification,artifacts and blurred details after defogging,this paper proposed a night fog-removal algorithm based on image layering to decompose the image into structural and texture layers.Defogging in the structural layer can retain the detailed information of the image to the greatest extent.In the texture layer,de-halo artifacts and noise reduction processing was conducted.Finally,the texture and structural layers were integrated to obtain the restored images.Experimental analysis and image quality evaluation show that the image restoration after night fog image restoration algorithm based on image layering has clear details and less noise.This thesis contains 23 figures,5 tables and 93 references.