基于双路权重映射与模糊核估计水下图像增强OA

To address the issue of image degradation caused by complex underwater environments,an underwater image enhancement algorithm was proposed.On one hand,the color corrected white balanced image was obtained by the red and blue channels compensation,and the white balanced image was divided into independent components which a gamma component and a sharpening component by adaptive contrast gamma correction and normalized unsharp masking,the normalized weight coefficients were derived through Laplacian contrast weight and saliency weight,which was designed to preserve more edge,texture,and structural details.On the other hand,Gaussian pyramid was introduced for normalized weight map sampling,meanwhile,Laplacian pyramid sampling was used for the dual-path image components.The initial reconstruction image was obtained based on a multi-scale pyramid fusion strategy.In addition,a variational energy equation is employed to reduce the blurring caused by scattering.The results show that compared with other state of art algorithms,the proposed method surpasses all advanced comparative algorithms in subjective assessment and objective assessment,which can effectively reduce color distortion and improve the image contrast and clarity.