Research On Image Denoising Method Based On Curvelet Transform And Total Variation

In the process of image transmission,noise is unavoidably contaminated.Therefore,image denoising has always been a hot issue in the field of image processing.Curvelet analysis is an image transformation domain description method based on multi-scale geometric analysis theory.It is based on the theory of wavelet and ridgelet analysis.It inherits the good time-frequency analysis capability of wavelet analysis and direction selection characteristics of ridgelet analysis.The Curvelet transform applied to the field of image denoising can better maintain the edge texture of the denoised image,but it will produce ringing effect;Total variation theory is a classical theory based on minimization of energy functional analysis.It smoothes the image during the process of variational problem solving.It does not cause image edge blur,but it also has staircase effect and makes edge texture details missing.Therefore,the research on the Curvelet transform and the full variational denoising theory has theoretical value for image denoising.First of all,this thesis proposes adaptive threshold selection for each Curvelet subband,and adopts a new threshold function between the soft and hard threshold functions.The difference from the traditional threshold function is that the function have two adaptive adjustment parameters,continuous and gradual characteristics,it also have simple expressions and flexible parameter selection.Secondly,this thesis proposes a method based on adaptive gradient fidelity total variation denoising.This method can adaptively select the regular term parameters and gradient fidelity term parameters to ensure the gradient between the noise image and the restored image.Finally,this thesis proposes an image fusion denoising algorithm based on Curvelet transform and total variation.This algorithm combines the advantages of the Curvelet transform with the total variational denoising method to overcome the shortcomings of Curvelet threshold denoising and the total variation denoising model.The fusion algorithm can well maintain important information such as the texture details of the edges of the image.The experimental simulation results show that the proposed algorithm not only effectively suppresses the image noise,improves the signal-to-noise ratio of the denoised image,but also preserves the edge texture details of the image better,and obtains better visual effects.