Research On Restoration Of Image In Foggy Weather

There is a lot of suspending particles in foggy weather. As a result of the scattering of these particles,an image acquired under this condition has poor contrast and is blurred meanwhile, with low contrast which make it very difficult for the image monitoring and recognizing, target tracking, aerocraft navigating and etc, so it is of great value for the research on restoration of an image captured on a foggy day. As the foggy image degrading is closely related to the scene depth, common image processing methods cannot achieve perfect enhancing and restoring effect.Up to the present, the research on this subject mainly focuses on the local histogram equalization and atmospheric scattering model. By in-depth analysis into atmospheric scattering model in this thesis, it is demonstrated that depth map based on the airlight model is unfeasible and meanwhile, simply blocking an image can not guarantee each block has the same scene depth, so local histogram equalization cannot achieve good restoration. As a result, this thesis will introduce another two algorithms of restoring the degraded images based on the depth.Local histogram equalization based on depth segmentation of scene is a algorithm, which combines the atmospheric scattering model with local histogram equalization one and depth map, as the basis of image blocking, applies to the local histogram equalization so that it gets better effect for the restoration of the single image. In this thesis, another algorithm is the atmospheric scattering model restoration based on the precise depth. As to the fixed spot, it can get the exact spot depth using two degraded images with great contrast. To avoid common algorithm which needs iterative normalized radiance, we use pre-estimated atmosphere scattering coefficient, get the sky brightness by histogram, accelerate the image restoration and guarantee its real-time. Simulating experiments prove the two algorithms both achieve satisfying effect.