Adaptive Underwater Image Enhancement Method Based On Image Fusion

The particularity of underwater imaging makes the light spreading in water affected by the absorption and scattering of particles in the water medium.This effect causes the light to be seriously attenuated,resulting in the underwater image directly collected underwater with poor visibility,blurry details,uncoordinated colors,low contrast,and noise,which seriously affect the reading of useful information in underwater images.In order to solve the above-mentioned common problems of underwater images,this paper propose an adaptive underwater image enhancement method based on image fusion.It takes the steps of color correction,brightness enhancement,and detail enhancement to process the original degraded underwater images to achieve the purpose of underwater images enhancement.The main contents of this article are as follows:(1)Aiming at the general color cast problem of underwater images,a color correction method based on color compensation is proposed.This method can restore the color cast of the underwater degraded image well,and it is better than other traditional white balance methods.(2)In order not to affect the color channel of the image after color correction,the image is converted to LAB space.Then an adaptive Gamma correction algorithm is used to adaptively adjust the brightness channel according to the light or dark characteristics of the image.The algorithm achieve the effect of balancing and adjusting the image brightness.(3)An adaptive underwater image enhancement method based on image fusion is proposed.First,the brightness channel of the color-corrected image and the brightness channel of the gamma-corrected image are used as the input images of wavelet decomposition.It also proposes to perform adaptive fusion based on L2 norm for the high-frequency components and linear fusion for low-frequency components.After reconstructing the fusion result of the luminance channel,combining it with the color channels,the final image enhancement result is obtained.(4)In view of the lack of reference images in the existing image data sets,the NYU data set is selected to establish composite blurred image data sets with different degrees of clarity according to the atmospheric scattering model and the attenuation coefficient settings.The establishment of the data set provides more abundant and referenced experimental data for the final enhanced method verification work.The specific experimental design is as follows.We first use the data set established in this paper and the non-reference data set downloaded on the network to conduct a comparison experiment of various methods.The experimental results show that compared with other methods,the proposed underwater image enhancement method is not only has good visual effects but also gets better results of image quality evaluation standards.In addition,a target recognition experiment based on YOLOv3 is designed,which proves that the method in this paper has a good effect in improving the number of recognition targets.