Research On Image Denoising Based On The Shearlet

Digital image has become an important means for people to obtain information and use information.However,the digital image in the acquisition,transmission,storage and other processes will be subject to different degrees of noise.In order to obtain the true image as possible,so as to meet the visual and image quality requirements,digital image denoising processing is necessary.Image denoising is one of the hot spots in digital image processing.The Gaussian noise in the image is one of the common noises,and wavelet transform is an effective denoising method for this kind of noise.In this paper,wavelet transform theory is deeply studied.Wavelet transform is a good time-frequency method,which is the crystallization of harmonic analysis.However,in recent years,it has been found that the wavelet transform is not sufficient for the representation of 2D images.Because the wavelet transform in the two-dimensional image of the sparseness of the image,the image direction of the lack of detail.In this paper,studied a new type of image representation method shearlet transform,which can realize multi-scale image segmentation through Laplacian pyramid and multi-direction image detail capture through pseudo-polar lattice.Based on the study of wavelet transform and shearlet transform,the following work has been done.Firstly,Several denoising functions in one-dimensional signals are introduced into two-dimensional image denoising.In this paper,wavelet threshold image denoising is studied,and several threshold functions are used to do the simulation under the VisuShrink threshold.Experimental results show that the improved soft threshold is better than other threshold functions.Secondly,Improved denoising of hard-threshold function images under shearlet transform.In this paper,the influence of shearlet decomposition scale and direction number on image denoising effect is studied.Through the wavelet threshold image denoising,the threshold function is applied to the shearlet transform denoising,the effect is not good.Therefore,the threshold function is improved to a certain extant.Finanly,An improved threshold function image denoising based on shearlet transform under unknown noise is proposed.In this paper,the wavelet threshold is used to estimate noise under the Bayesian frame.Then the estimated noise used to the shearlet transform denoising through the hard-threshold and improved-threshold.The effectiveness of the proposed method is proved by the simulation result.