Research Of Underwater Image Enhancement And Restoration Methods

In recent years,with the reduction of land resources,marine resources have become the focus of continuous struggle.Marine images can directly reflect underwater environment information,so underwater image analysis and processing is conducive to the research of marine exploration,marine environmental monitoring,marine military applications.In order to efficiently obtain marine information,it is important to use image processing technology to clarify underwater images.In this context,the main work of this paper has the following points:Firstly,underwater images are analyzed based on non-physical model and physical model respectively by analyzing the absorption and scattering of light during underwater transmission,which provides the basis for subsequent underwater image enhancement and restoration.Secondly,an underwater image enhancement algorithm based on hierarchical weighted fusion is proposed.Guided image filtering is taken into consideration which can effectively remove noise and retain edge information.Then Laplacian Pyramid is used to decompose the filtered image for subsequent image fusion.In addition,the proposed method introduces weight influence factor to modify weight maps which aims to further enhance edge.Finally,the enhanced images are obtained by reconstructed function of Laplacian Pyramid.The proposed algorithm improves contrast and the color is corrected.Thirdly,a method of restoring underwater image based on convolution neural network is proposed.An underwater image dataset is established by simulating underwater image formatting process by using clear images and underwater degraded model.Then,this paper proposes a convolution neural network model for underwater image enhancement and the model can directly create a mapping relationship between clear image and underwater image.Finally,underwater images are restored by the features extracted by convolution neural network.The images obtained by proposed method improve the clarity and the color is corrected,which provides a way for the restoration of underwater images.