Dehazing method of underground pipeline image based on dark channel and retinex

Abstract

The invention relates to a method for defogging underground pipeline images based on dark channels and Retinex. By preprocessing the image of the foggy underground pipeline, the dark channel image corresponding to the foggy image is obtained; mean filtering is performed on the obtained dark channel image to approximate the image transmittance; an offset value is compensated for the result of the mean filtering, and the obtained Rough estimate of transmittance; Calculate the expression of estimated transmittance; Use the pixel values ​​of the original image and the mean-filtered image to estimate the global atmospheric light; Use the physical restoration model to restore the preliminary defogged image; Use the multi-scale Retinex algorithm Enhance the preliminary defogged image; add a color restoration factor to compensate for the local distortion of the color; add dynamic range adjustment parameters to control the color cast of the image; Enhanced recovery and more. It can be used in scientific fields such as video monitoring of underground pipeline environment, underground pipeline disease identification, etc.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a method for defogging underground pipeline images based on dark channels and Retinex. Background technique [0002] In the urban underground pipeline environment, the internal image of the pipeline acquired by the pipeline disease detection robot plays an important role in the identification of pipeline diseases. The image of the pipeline received has degradation phenomena such as low contrast and fuzzy details, which affect the normal work of the vision system of the pipeline inspection robot. Therefore, dehazing and restoring the collected images is of great significance for pipeline safety. [0003] Most of the current dehazing methods are for foggy images of outdoor scenes, and there are few algorithms for foggy images of pipeline environments. Researchers mainly study the problem of image dehazing from two directions: the direction based on image enhancement and ...

AI Technical Summary

Problems solved by technology

However, urban underground pipelines are in a dark or low-light environment, and the collection of pipeline internal images by pipeline robots depends on their own light sources, which also leads to the collection of pipeline images usually with large reflective areas, making the original dark primary colors The dehazing effect of the dehazing algorithm in the underground pipeline scene is not good

In addition, according to the image restoration algorithm based on the dark channel prior based on the atmospheric illumination model, the color of the image is often unbalanced, and the restored image is overall dark.

Images