Dark channel prior based image defogging method

Abstract

The invention discloses a dark channel prior based image defogging method which includes the steps of 1, dividing the image I(x) on the basis of the classical clustering algorithm to acquire a candidate sky region It(x); 2, corroding the candidate sky region It(x) to acquire sky brightness; 3, performing mini-value filtering on the image to improve the same according to the sky brightness so as to acquire a coarse-estimation transmissivity image; 4, optimizing the coarse-estimation transmissivity image by means of guide filtering to acquire a transmissivity image; and 5, acquiring a recovered image based on an atmospheric scattering model. The dark channel prior based image defogging method has the advantages of processing airglow more reasonably, reducing image block effect, improving overall brightness of images and making the images more natural.

Description

technical field [0001] The invention belongs to the field of image processing, and in particular relates to an image defogging method based on dark channel priors, which can perform defogging processing on images. Background technique [0002] Recently, haze weather has become a frequent weather phenomenon. The liquid droplets and small solid particles in the smog not only endanger human health, but also the scattering effect of a large number of suspended particles reduces the visibility of the atmosphere, reduces the image quality, and affects the acquisition of relevant information in the image. At the same time, the reduction of outdoor clarity also leads to unnecessary Traffic accidents, therefore, image defogging technology has become one of the urgently needed research topics in the field of image processing and computer vision. [0003] Image defogging techniques mainly include two types based on image restoration and image enhancement. Based on the image restorati...

AI Technical Summary

Problems solved by technology

Gibson et al. proposed to use median filtering instead of minimum filtering, which can weaken the Halo effect. This algorithm does not require soft matting or bilateral filtering to refine the transmittance map, which significantly reduces the time complexity of the operation. However, this method removes haze images. Poor quality, prone to black spot effect

Zhang Xiaogang proposed dual-area filtering, defined the dark area, improved the median filtering algorithm, and weakened its black spot effect, but the algorithm has limited effect on maintaining image details, and the fineness of image details is reduced

[0009] The above literatures all take the atmospheric light as a constant, which is obviously unreasonable in images with changing depth of field; in addition, the algorithms based on the dark channel prior all have dark gray images with Halo effect, or to eliminate the Halo effect However, complex calculations increase the time complexity of the algorithm

Images