Image defogging algorithm on basis of dark channel prior

Abstract

The invention discloses an image defogging algorithm on the basis of dark channel prior. The image defogging algorithm includes steps of 1, solving dark channel graphs of foggy degraded images; 2, solving rough transmittance of the images by the aid of the dark channel graphs; 3, carrying out amplification correction on transmittance of bright regions such as sky by the aid of improved tolerance mechanism transmittance adjusting algorithms on the basis of color attenuation prior to obtain corrected transmittance; 4, refining the transmittance by means of guide wave filtering; 5, combining halation operators and the dark channel graphs with one another to acquire atmospheric light values; 6, acquiring restored images J by the aid of atmospheric scattering models. The image defogging algorithm has the advantages that the problem of color cast of bright regions such as sky in restored images due to the traditional defogging algorithms can be effectively solved by the aid of the image defogging algorithm; the problem that the traditional methods for solving atmospheric light values are vulnerable to disturbance of white objects can be solved by the aid of the image defogging algorithm;the quality of the images can be improved by the image defogging algorithm.

Description

technical field [0001] The invention relates to the field of image processing, in particular to an image defogging algorithm based on dark channel prior. Background technique [0002] In the case of haze weather, the images acquired by the vision system will be affected by the scattering of visible light by haze particles, resulting in degradation phenomena such as low contrast and blurred details, which will affect the normal operation of the vision system. Therefore, image restoration in foggy weather is an urgent research work. At present, the image defogging algorithm mainly has two directions: model-based dehazing and non-model-based dehazing. The essence of non-model-based dehazing is to enhance image contrast, but this type of method may lose image information and cause image distortion, which is represented by the Retinex algorithm. The essence of the model-based dehazing method is to construct a physical model of the foggy image, and reverse the degradation proces...

AI Technical Summary

Problems solved by technology

Based on the prior knowledge that the brightness and saturation of foggy image pixels change sharply with the change of haze concentration, Zhu et al. constructed a linear model to obtain the depth information of the image, so as to realize image defogging, but the algorithm involved many parameters. , poor adaptability

Wang et al. proposed to adjust the transmittance of the sky region by identifying and extracting the sky region by setting the gradient and brightness threshold according to the characteristics of the sky region. However, this algorithm uses the largest connected region as the sky for recognition, which is time-consuming and easy. Missing part of the sky, resulting in color distortion in the missing area

Jiang Jianguo introduced a tolerance mechanism to judge bright areas such as the sky, and improved the transmittance of the area, which better solved the problem of color distortion in the sky area and became a mainstream solution. The difference parameter selection is affected by whether there is a sky area in the image, and the transmittance after correction will be insufficiently enhanced or overcompensated

Images