Image Dehazing Method Fused with Variation Function and Color Attenuation Prior

Abstract

The invention discloses an image defogging method that combines variation function and color attenuation prior. Firstly, the traffic image under foggy weather is obtained; Attenuate the sky area under the priori, and use the variogram to judge the sky area of ​​the image; then calculate the image atmospheric ambient light value and scene transmittance according to the sky area judged by the two; finally process the image according to the atmospheric scattering model , and then use the automatic color scale method to correct the color of the image, and output the processed fogged degraded image. Compared with the traditional method, the present invention has great advantages in selecting the atmospheric ambient light value and scene transmittance of the image, and the image distortion after restoration is small, which is of great significance to the further processing of the image and the accurate acquisition of image information.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to an image defogging method by fusing variation functions and color attenuation priors. Background technique [0002] In recent years, with the continuous development of the field of image processing, using images to obtain information has become an important means. All outdoor computer vision systems, including automatic surveillance, intelligent vehicles, and outdoor object recognition, require high-quality images, however, images captured in harsh weather conditions are usually scattered by a large number of suspended particles in the atmosphere, causing capture Due to the existence of a large amount of noise, it is difficult for the computer vision system to obtain accurate information based on such low-quality images. [0003] In recent years, the dehazing algorithm based on a single image without other additional information has made great progress. At ...

AI Technical Summary

Problems solved by technology

.Usually, the latter's dehazing effect is better than the former two, because the former two have not considered the degradation mechanism of the foggy image, while the latter is based on the degraded model, using prior knowledge for reverse restoration, for Most images can remove fog well, but these three types of algorithms have their own limitations, and their robustness is not strong

For example, Tan maximizes the contrast of its neighborhood to calculate the optimal transmittance combination of each pixel, and uses the Markov random field to determine the optimal transmittance in this combination, which is still a contrast stretching algorithm in essence. The color of the restored image is too bright; Nishino et al. use the Bayesian posterior probability model to fully mine the potential statistical features in the image for dehazing processing. This method can handle dense fog well, but when dealing with thin fog, its color It is also too bright and lacks a sense of reality; Fattal assumes that the chromaticity and transmittance in the neighborhood are not correlated, and dehazes according to the statistical characteristics of the color, so the effect of processing areas with inconspicuous features is poor; Tarel et al. use the median filter to estimate Dissipation function, and the median filter does not have the edge-preserving feature, resulting in a small amount of fog remaining in the restored image at the sudden change in depth; He et al. first proposed the prior knowledge of the black channel, using this prior knowledge can obtain a good dehazing effect, but Due to the extremely high space-time complexity of its matting processing, the algorithm is not real-time. In recent years, many researchers have proposed corresponding improvement strategies for the He algorithm. Although the dehazing efficiency has been substantially improved, but All at the cost of sacrificing the quality of defogging

Almost all the dehazing algorithms based on the scattering model will more or less have over-enhancement when dealing with the sky area, resulting in poor subjective visual quality of the restored image

Images