Research On Underwater Degradation Image Restoration Method Based On Polarization Detection

When acquiring images in the natural environment,the quality of the acquired images is severely degraded due to the complex and changeable underwater environment.The main reason is the absorption and scattering of light by suspended particles,which causes the attenuation of light.Compared with the visible light image,the polarized image can more prominently highlight the edge contour of the target,thereby improving the use value of the image.The main research of this paper is divided into two aspects,namely the experimental design of the acquisition of polarized images and the research of the restoration methods of degraded images.This article analyzes the related factors of underwater image degradation.Aiming at the problems of poor underwater image quality,blurred texture details and low contrast,a set of underwater polarization imaging experiments were designed to study the imaging characteristics of polarized images under different turbidity levels.By configuring milk solutions of different concentrations to simulate an underwater turbid environment,the SALSA camera was used to obtain underwater polarized images of different materials and targets at different concentrations under the illumination of natural light,and acquired the polarized intensity images at different angles The Stokes vector method is used to obtain polarization components such as polarization degree and polarization angle.Because the polarized image has strong edge information and texture information,and the intensity image has a lot of light intensity information,this paper proposes a compressed sensing image fusion algorithm based on NSCT transform.The polarization image and intensity image are used as the source image.The transformation generates low-frequency components that reflect approximate information and high-frequency components that reflect detailed information.The low-frequency coefficients are based on a fusion method based on the area energy as a weight ratio.The high-frequency coefficients are based on adaptive selection and compression based on the combination of double threshold spatial frequency and correlation coefficient.A fusion optimization algorithm combining perceptual thoughts to achieve the purpose of polarized image restoration.Experimental results show that the algorithm proposed in this paper can reconstruct higher quality images with less data.Compared with traditional image restoration algorithms,traditional image fusion algorithms,and traditional compressed sensing reconstruction algorithms,the algorithm in this paper has certain advantages in improving image evaluation indexes such as image standard deviation and information entropy.