Single-image defogging method based on dual-channel prior and side window guided filtering

Abstract

The invention discloses a single-image defogging method based on dual-channel prior and side window guided filtering, and the method comprises the following steps: (1) respectively solving a dark channel image and a bright channel image of a foggy image, and respectively taking the first 0.1% of pixel points with the maximum gray values of the two images; (2) solving a self-adaptive threshold delta (0<delta<1) by utilizing the brightest pixel value and the darkest pixel value design function in the dark channel image and the bright channel image of the original fog image; (3) solving an atmospheric light value A by utilizing the distribution of pixels in the RGB three-dimensional space diagram decision image proposed by the literature [6]; (4) obtaining transmissivity according to dark channel prior; (5) combining side window filtering and guiding filtering to form side window guiding filtering; and (6) according to an atmospheric scattering model, substituting the atmospheric light value A and the transmissivity into an optical expression of a foggy day image imaging model to obtain a final restored scene.

Description

technical field [0001] The invention belongs to the field of computer image processing and is used in related fields such as image or video defogging; in particular, it relates to a single image defogging method based on dual-channel prior and side window-guided filtering. Background technique [0002] During the image acquisition process, due to the influence of fog, the visibility of the scene is greatly reduced, coupled with the influence of atmospheric light, the obtained image is seriously degraded in quality and the contrast is greatly reduced. The degradation of image quality will seriously affect its subsequent processing, such as image comparison, feature extraction, image recognition, etc. Image defogging has always been a hot spot in the field of computer vision and image processing. [0003] Among the existing traditional image dehazing methods, He et al. [2] The dehazing method based on dark channel prior has been widely used, but the soft matting algorithm us...

AI Technical Summary

Problems solved by technology

[0003] Among the existing traditional image dehazing methods, He et al. [2] The dehazing method based on dark channel prior has been widely used, but the soft matting algorithm it uses is too complicated, which greatly increases the amount of calculation

Although many subsequent methods have improved it and achieved good defogging effects, there are still problems such as incomplete defogging of edge mutation areas; when a large number of white areas appear in the image, the restored image effect is not good, etc.

Images