Improvement Of Total Variational Image Denoising Algorithm Based On Compressed Sensing

Image acquisition,transmission and storage process will inevitably be affected by noise.In order to get a clear image with higher resolution,it is necessary to denoise the image containing noise.The denoising of image containing noise has become a basic problem in image processing.Compressed sensing,as a new signal sampling and processing theory,breaks through the limitation of the Nyquist sampling theorem and samples at the same time of signal compression,which relives the pressure of signal sampling and storage.It has become one of the research hotspots of current signal processing.Considering that both the reconstruction algorithm based on compressed sensing and the denoising method based on partial differential equation have certain deficiencies,this paper focuses on the image denoising theory based on compressed sensing,which is mainly divided into the following two aspects:(1)the image denoising algorithm based on partial differential equation(PDEs)is described,and the Perona-Malik model based on partial differential equation,the restoration model based on fourth order partial differential equation and the classical total variation model are mainly introduced.In different test images and noise variances,the denoising experiments of the above algorithms were carried out.The denoising effects of the different algorithms were measured by the peak signal-to-noise ratio(PSNR)and structural similarity index,and the denoising characteristics of the different algorithms were analyzed.Aiming at the disadvantage that the image is too smooth after the denoising of the total variation model,the improved total variation denoising algorithm was proposed by introducing edge guidance function and designing a new model.The compressed sensing theory was combined with the idea of total variation denoising.The experimental results show that,in terms of the PSNR and structural similarity index,the improved model proposed in this paper reduces the degree of over-smoothing of images after denoising,and the PSNR is higher.(2)In view of the phenomenon that the restored image obtained by SL0 reconstruction algorithm based on contourlet transform has lost details such as edge and contour,the contourlet transform is combined with the improved total variation image denoising algorithm.The experimental results show that contourlet can effectively capture the contour of the original image,and is better than discrete cosine transform and discrete wavelet transform in feature extraction of image texture and shape.The improved total variation algorithm can alleviate the loss of image edge details.Experimental results show that the improved model can effectively improve the quality of image reconstruction.