Fully Adaptive Denoising Algorithm Based On Improved Adaptive TV Model And K-SVD Model

The main contents of this paper are the total variation model,the application of K-SVD and wavelet transform in de-noising,the adaptive selection of regular term coefficients,and the optimization and improvement of related algorithms.Nowadays,with the continuous progress and development of digital technology,digital image processing has been more and more widely applied in production and life,such as biometric identification,traffic management,metallurgy and heavy industry,military and other fields.It has gradually become an indispensable and powerful tool in the development of smart city and smart society.In the process of digital image processing,it is inevitable for images to be interfered by a variety of factors,so the degradation of image quality and even the loss of its research value because of noise pollution will inevitably occur.Therefore,it is particularly important to improve the technology of image processing so as to effectively reduce the influence of noise and preserve the research value of important image data.In the current image de-noising model,the total variation image restoration model has attracted extensive attention in the field of image processing due to its good mathematical properties,efficient numerical algorithm and relatively few limitations on the image itself.Wavelet transforms,with its advantages of zoom,can make the image more efficient in the frequency domain to store the information of edge and flat region,and also effectively help this paper to obtain more accurate flat and edge graphs when preprocessing the image.As the most concerned image processing model in the past 10 years,K-SVD model plays a great role in promoting the field of image de-noising because it can process images more quickly and at the same time has strong de-noising effect.This paper first introduces several classical total variation model,and mainly studies the adaptive de-noising total variation model,according to the advantages and disadvantages of the model,the total variation adaptive de-noising model is improved and established the hydrolysis of fidelity term equivalent to flat figure fidelity term and the new model of the fidelity term marginal figure,including flat image and edge image acquisition by wavelet transforming.Then,the ROF algorithm for solving the total variation de-noising model is introduced.After in-depth study,the ROF algorithm is optimized so that the regular term coefficients manually selected can be selected by computer adaptively.Finally,this paper lists and introduce the k-svd model,and regularizes the K-SVD model to make it adapt to the main research content of the text--adaptive total variation de-noising model.Thus to an accurate and convenient image is the de-noising model.Through the experimental verification,the proposed model and algorithm inherit the advantages of the total variation model,and have a good experimental effect on edge protection and noise removal.At the same time,it simplifies the experiment process(cancels the steps of doing a lot of experiments to select appropriate regular term coefficients),and,because of the integration of K-SVD model,greatly reduces the running time of the computer and the time cost.