Defogging method for image containing sky area

Abstract

The invention discloses a defogging method for an image containing a sky area, and belongs to the technical field of image processing. The method mainly comprises the following steps: S1, acquiring an image I to be processed, and segmenting the sky and other parts by using a weak supervision image segmentation method which mainly comprises the following steps: making graffiti semantic segmentation tags, establishing cluster prediction tags and establishing a loss function; s2, carrying out defogging processing on other segmented parts by using an improved dark channel method, wherein the defogging processing mainly comprises the following steps: designing a defogging model and a transmissivity model, adding fast steering filtering and setting an atmospheric light value; s3, carrying out defogging processing on the sky area by using an MSRCR method; and S4, splicing the defogged sky area and other parts into an image. The algorithm is utilized to effectively solve the problem of artifacts generated during defogging of a sky-containing area, and compared with a defogging method using deep learning, the defogging time is greatly shortened.

Description

technical field [0001] The invention belongs to the technical field of deep learning and computer vision, and in particular relates to a method for defogging an image of a sky region. Background technique [0002] Electric energy is indispensable to human life. From the perspective of power generation composition, thermal power generation in various countries occupies a dominant position, but fossil fuels in thermal power generation, including coal, oil, and natural gas, generate a lot of pollution during the combustion process, which aggravates the The harm of acid rain also increases the dust pollution in the air, which has a serious impact on people's life and the survival of plants, and its resource consumption is large, the efficiency is low, aggravating the shortage of resources, and causing many troubles for environmental governance. Compared with wind power, which is one of the renewable energy sources, its advantages are particularly prominent. [0003] The windwar...

AI Technical Summary

Problems solved by technology

However, when the dark channel dehazing algorithm faces bright object scenes such as sky areas or white buildings, halos appear due to inaccurate estimation of atmospheric light values; dehazing algorithms based on deep learning often require manual training Each pixel in the set image defines a label. Although the final effect is remarkable, the workload is huge and the time is long.

Images