An underwater image sharpening processing method based on multi-scale gradient domain contrast stretching

Abstract

The invention discloses an underwater image sharpening processing method based on multi-scale gradient domain contrast stretching, and the method comprises the steps: employing a white balance algorithm to obtain an underwater image with a blue window effect removed, and converting the underwater image from an RGB color space into an LAB color space; decomposing the L-channel image of the LAB color space of the underwater image based on a weighted least square method to obtain a rough image; calculating a detail image according to the rough image, and enhancing the detail image by adopting different gradients to obtain an enhanced detail image; carrying out gamma correction on the rough image, and using the corrected rough image as a basic image; synthesizing the enhanced detail image andthe basic image to obtain an image after contrast enhancement; carrying out color saturation compensation on channels A and B of the LAB color space of the underwater image; and converting the image after contrast enhancement and the underwater image after color saturation compensation into RGB color space. According to the invention, the definition of the underwater image is improved.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to an underwater image clarity processing method based on multi-scale gradient domain contrast stretching. Background technique [0002] With the development of image processing technology, underwater images are clearer, and the optical vision system of underwater robots is embedded in the form of algorithm modules to improve the visibility of underwater turbid images and facilitate accurate target recognition in the later stage. Clear images in foggy days on the road and underwater images, embedded in locomotive remote observation systems, small car driving assistance systems and driving recorders, and underwater robot optical vision systems in the form of algorithm modules, to improve foggy images or underwater turbid images Medium visibility, which is convenient for accurate target identification in the later stage. [0003] Multi-parameter estimation ill-posed problem:...

AI Technical Summary

Problems solved by technology

[0003] Multi-parameter estimation ill-posed problem: At present, underwater image restoration based on physical model is essentially a problem of solving ill-conditioned equations, and it is an incompletely constrained problem

Assumptions and prior knowledge are limited, and the processing effect depends on the estimation accuracy of multiple parameters, which is prone to distortion. It is only practical for underwater images or certain degradation phenomena under specific conditions. In the face of real and complex oceans environment, its comprehensiveness, robustness and accuracy are not ideal, and it is severely limited in practical applications

Noise amplification problem: The marine environment contains a large amount of sediment, dissolved organic matter and tiny suspended particles, resulting in rich noise in underwater images, and also amplifies the impact of backscattering on underwater imaging

Especially in the deep sea environment, a large number of floating "marine snow" causes target occlusion, which seriously damages the underwater imaging quality

At present, a large number of studies have effectively improved the visibility of underwater images, but also seriously amplified the noise, making the target of interest submerged in the noise, resulting in further deterioration of the underwater image quality.

[0004] In summary, the clarity of underwater image processing results needs to be improved

Images