Sky region segmentation-based single image defogging method

Abstract

The invention discloses a sky region segmentation-based single image defogging method, and overcomes the shortcoming of a negative visual effect due to the fact that a conventional dark channel defogging algorithm is not adaptive to a sky region. By comprehensively utilizing visual features in an image, the sky region in the image is accurately segmented out by adopting a simple linear iterative clustering and edge detection combined method. Based on this, a non-sky region is subjected to independent transmissivity estimation according to average saturation priori, and the sky region is subjected to independent visual effect enhancement processing, so that the negative effects of over-enhancement, color cast and the like caused by performing conventional defogging processing on the sky region are eliminated. Finally, for the shortcoming that the defogged image is relatively dark, a steerable filter-based Retinex method is proposed for performing enhancement, so that the visual effect of the defogged image is further improved.

Description

technical field [0001] The invention relates to a single image defogging method based on sky area segmentation, and belongs to the technical field of image processing. Background technique [0002] The images / videos captured in hazy weather are affected by the absorption, scattering and refraction of light by turbid media in the atmosphere (such as hydrosol, fog, haze, etc.) This results in reduced image visibility, such as blurred contrast, color migration, and reduced dynamic range. Given that many applications in the fields of image understanding, target recognition, and target tracking in computer vision systems are based on the premise that the input image or video is shot under ideal weather conditions, the clarity of the image is critical to subsequent interpretation, analysis, recognition, and measurement results. accuracy will have a certain impact. Therefore, the study of fast image dehazing algorithm has attracted more and more attention of researchers. [0003...

AI Technical Summary

Problems solved by technology

[0003] In recent years, single image defogging has attracted the attention of many scholars. These methods realize image defogging by using the prior information contained in a single image or proposing some reasonable assumptions. He et al. through a large number of outdoor haze-free images According to the observation, a dark channel prior defogging algorithm is proposed, which has achieved good results, but there are still defects

The soft matting algorithm adopted by this method when refining the transmittance is complex, and it is difficult to achieve real-time dehazing of the image; since the sky region does not satisfy the dark channel prior, the restored image is prone to noise and color distortion in the sky region ;In addition, the color of the image after defogging through the dark channel is dark, which reduces some visual effects

Images