Research On Underwater Image Enhancement Method Based On Local Correction And Multi-scale Fusion

Information analysis of underwater optical images is one of the important means to obtain marine information,which is of great significance in the fields of marine exploration,marine resource development,and archaeology.However,due to the physical properties of the underwater environment,underwater images usually show blur,low contrast,and serious color cast.To solve these problems,this dissertation proposes an underwater image enhancement method based on local contrast correction and multi-scale fusion.The specific research content is divided into the following three parts:Firstly,aiming at the problem of color deviation of underwater image,considering that the pixel value of the red channel is usually low,this dissertation uses the method based on red channel compensation to supplement the pixels of the image,and then carries out white balance processing on the compensated image,which can effectively alleviate the oversaturation phenomenon of the red channel caused by the direct use of white balance processing.Secondly,the white balance processed image is further enhanced to form two different images with local contrast correction and sharpening:(1)Aiming at the uneven illumination of underwater images,a local contrast correction method based on guided filter mask is proposed.This method simultaneously performs suppression and enhancement operations on the over-bright and over-dark areas of the image,which can effectively improve the influence of uneven illumination on the collection of underwater images information and obtain a contrast-corrected input image.(2)Aiming at the problem that the details of the underwater image are not obvious,the guided filter is used to sharpen the image to obtain an input image with enhanced details.These two different images perform effective contrast correction and detail enhancement on the white-balanced image respectively,and provide input images for the fusion strategy.Finally,calculate the weights of the above images,select the Laplacian contrast weight,saliency weight,and saturation weight,and calculate the normalized weights of the two sets of weight maps.Using a multi-scale fusion strategy,the input image and the normalized weight image are decomposed by Laplacian pyramid and Gaussian pyramid respectively,and the final enhanced image is obtained through fusion and reconstruction.This dissertation systematically researches underwater image enhancement methods based on local contrast correction and multi-scale fusion.Through the research of the underwater environment,the problems of color distortion,low contrast and unobvious details of underwater images have been effectively solved.At the same time,this dissertation provides important theoretical and practical reference value for the future development of marine environment exploration,marine resource development and archaeology.