Real-time video de-fogging method using gray-scale image guided filtering

Abstract

The invention discloses a real-time video de-fogging method using gray-scale image guided filtering. The method comprises the following steps that: an image is obtained in a video stream; the original image is converted into a gray-scale image; guided filtering is performed on the gray-scale image; downsampling is performed on the original image, so that a thumbnail is obtained; the dark primary color image of the thumbnail is obtained according to a dark primary color definition formula; a quadtree method is adopted to find a dense fog area or a sky area; and an improved atmospheric light scattering model is used to restore an image, and a foggy day video is outputted. With the method adopted, the defogging effect of the restored image is good, and the colors of the restored image are bright and vivid; problems of white edge phenomena occurring in places where depth of field changes greatly, and color distortion in a sky area which is caused by a dark primary color theory-based de-fogging algorithm can be solved. The method has a very good de-fogging effect for light fog, moderate fog and dense fog; the computation amount of the method is small; and the processing speed of the method is high and stable.

Description

technical field [0001] The invention relates to the technical field of video processing, in particular to a real-time video defogging method using grayscale image-guided filtering. Background technique [0002] In foggy weather conditions, due to the light scattering effect of suspended particles in the atmosphere, the contrast and visibility of the captured video are reduced, which seriously affects the normal work of the outdoor image acquisition system. Therefore, it is very meaningful to perform real-time defogging processing on videos captured in foggy weather to improve video quality. [0003] In recent years, scholars at home and abroad have mainly studied the dehazing of a single image, and have made great breakthroughs in this field. At present, image processing methods for foggy days are mainly divided into two categories: image enhancement-based dehazing algorithms and physical model-based image restoration and dehazing algorithms. Typical image enhancement meth...

AI Technical Summary

Problems solved by technology

Typical image enhancement methods such as the defogging algorithm based on Retinex theory can effectively increase contrast, highlight details, and can defog and improve the visual effect of foggy images to a certain extent. However, since the degradation of the image is not considered, The ability to restore image information is limited, which can easily lead to the loss of part of the image information, and the haze cannot be fundamentally removed, and the processed image is prone to grayscale, and the algorithm complexity is high, which cannot meet the real-time requirements

The other is a typical method of image restoration based on a physical model, such as the dark channel prior defogging algorithm. In the dark channel prior defogging algorithm, the atmospheric light in the atmospheric scattering model is assumed to be a constant value, and the transmittance is calculated through the dark channel prior theory. Figure to restore foggy images, for general outdoor foggy images can achieve better defogging effect, but the complexity of the algorithm is not as high as the requirements of real-time video processing, and the sky area of ​​the image after defogging is prone to oversaturation and color distortion , the overall image is too dark, etc.

Scholars at home and abroad have proposed many defogging algorithms, but most of them are only applicable to the defogging of a single image. At present, the research results of real-time video defogging are still relatively few, and most of the existing methods are in terms of computing speed and The defogging effect is still unsatisfactory, and the processing speed and defogging effect cannot be considered

Images