Image defogging method based on prior information

Abstract

The invention discloses an image defogging method based on prior information. The method comprises the following steps: estimating an atmospheric light value according to dark channel prior in combination with distant view dense fog area pixel points; respectively determining atmospheric transmissivity corresponding to distant view dense fog area pixel points and close view thin fog area pixel points according to color attenuation priori; determining atmospheric transmissivity corresponding to pixel points in the close-range mist area according to dark channel prior; determining pixel values corresponding to pixel points of the defogged distant view dense fog area and pixel values corresponding to pixel points of the defogged close view thin fog area; and performing region combination on the pixel values corresponding to the defogged distant view dense fog region pixel points and the defogged close view thin fog region pixel points to obtain a defogged image. By improving the defects of dark channel prior and color attenuation prior defogging algorithms in defogging application, the defogging method has a good defogging effect on mist images in various scenes.

Description

Technical field [0001] The invention belongs to the technical field of image processing, and specifically relates to an image defogging method based on prior information. Background technique [0002] At present, defogging algorithms based on image restoration research fog image imaging models, substitute known parameters and solve unknown parameters to achieve defogging, and are mainly divided into two methods: prior knowledge and machine learning. [0003] Traditional defogging methods are mainly based on prior knowledge, including He Kaiming et al.'s dark channel prior method, Tan et al.'s maximum contrast method, Zhu et al.'s color attenuation prior method, Ancuti et al.'s chromaticity inconsistency method, etc. . A prerequisite for using the dark channel attenuation prior algorithm is "non-sky". Since the pixel value of each color channel in the sky area is very high, there is no dark channel value tending to 0, so the dark channel is used to remove the fog a priori , It is ...

AI Technical Summary

Problems solved by technology

Then, the DehazeNet end-to-end system was proposed for transmittance estimation. Ren et al. successively proposed MSCNN and GFN to realize the dehazing of a single foggy image; however, due to insufficient data, lack of real haze-free images, and no direct For problems such as dehazing processing, the image dehazing algorithm based on deep learning is still in the development stage, and there are still many deficiencies.

Images