A Retinex-Based Fast Image Restoration Method in Nighttime Fog

Abstract

The present invention relates to a Retinex-based fast nighttime fog image restoration method. According to the method, a single color image of nighttime fog is processed, so that the visual perception for the color image can be clear; the illuminance images of each color component are estimated through adopting a guided filter with an edge-preserving property, and local contrast coefficients are calculated; Otsu threshold segmentation is performed on a maximum value channel image of the original image; the Gamma parameters of each pixel are estimated adaptively; Gamma correction and contrast preservation are carried out on the illuminance images; a brightness enhanced image is transformed into an HSV space; contrast enhancement is carried out on a brightness channel V through a CLAHE algorithm; and the image is transformed into an RGB space, so that a restored image can be obtained. The method of the invention has universality. With the method adopted, the definition of a defogged image can be effectively improved, the color cast phenomenon of the image can be corrected effectively, the details of the enhanced image are clear, and the colors of the enhanced image are natural.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a Retinex-based fast image restoration method in foggy days at night. Background technique [0002] Fog is a natural phenomenon, which is caused by a large number of tiny water droplets, aerosols and other media suspended in the atmosphere, so that the light reflected from the surface of the object and reaching the camera is attenuated by the scattering of atmospheric particles, so that the contrast of the image collected by the imaging device , desaturation or even hue shift. Under low illumination conditions, if affected by fog, the collected images tend to have low brightness, low contrast and color cast, making image details unrecognizable, which seriously affects the visual effect of the image and the analysis and analysis of the image by the computer vision system. Therefore, it is necessary to enhance the foggy image at night to restore a clear image. [0003] At present,...

AI Technical Summary

Problems solved by technology

This method can highlight image details, but due to the use of global color transformation without considering the image information under different scenes and lighting conditions, there is a color cast in the enhanced results

Another researcher proposes to perform illumination compensation and color correction on the image first, and then use an improved dark channel prior-based dehazing algorithm to enhance the foggy image at night. This method can effectively enhance the brightness and contrast of the image, and overcome the Color cast phenomenon, but due to the dehazing method based on the dark channel prior, the enhanced image sky is prone to block effects

Traditional image defogging methods can no longer meet the restoration requirements of nighttime foggy images, so it is necessary to combine the characteristics of nighttime foggy images to restore the nighttime foggy images, so as to improve the visual effect of the image and facilitate the analysis of the image by the computer vision system and processing

Images