Research On Image Denoising Algorithm Based On Diffusion Model And Wave Domain Transformatio

As the information dissemination carrier of modern society,images are an indispensable part of scientific and technological life.However,some noises are easily generated in the process of image acquisition and transmission,and these noises will affect the acquisition of information.Therefore,effectively preserving important information while removing image noise is an indispensable part in the field of image processing research.This paper firstly proposes an image denoising algorithm based on Tetrolet transform and BM3 D to solve the problem that the classical BM3 D algorithm has unsatisfactory performance in high-intensity noise denoising and will lose details such as image edge texture.Then,in view of the lack of information such as fuzzy edge details in the anisotropic diffusion model,an adaptive image denoising algorithm based on local variance and wave domain Lo G operator is proposed,and the gradient information is used in this new method to characterize the local area.Determining the degree of diffusion by features has its limitations.Finally,a wave domain image denoising algorithm based on Gaussian curvature and LMS algorithm is proposed.The main work is as follows:1.An image denoising algorithm based on Tetrolet transform and BM3 D is proposed.The new algorithm uses Tetrolet transform to pre-filter high-noise images to reduce image noise,protect the internal structure and details of the image as much as possible,and combine the Sobel operator for edge detection.After dividing the image into non-edge areas and edge areas,use BM3 D Denoising and searching in different directions can speed up the search for similar blocks and increase the number of similar blocks.Finally,combining structural similarity with Euclidean distance makes the grouping of image blocks more accurate.The simulation results show that the new algorithm has a higher peak signal-to-noise ratio,and the details of the image are more preserved.2.An adaptive image denoising algorithm based on local variance and wave domain Lo G operator is proposed.The new algorithm decomposes the image through wavelet transform to extract high-frequency parts,uses the Lo G operator in the enhanced control function to detect the edge texture,combines the local variance and image gradient to control the diffusion intensity,and selects the diffusion model adaptively according to the control parameters.Diffusion denoising.The simulation results show that the new model can not only effectively remove the noise in the image,but also keep the details such as edge texture intact.3.A wave-domain image denoising algorithm based on Gaussian curvature and LMS algorithm is proposed.Firstly,using the geometric properties of the image,the Gaussian curvature is introduced into the diffusion model as a detection operator,and then the highfrequency part of the image is extracted by the wavelet transform.Diffusion intensity achieves denoising.The simulation results show that the algorithm can effectively balance the removal of image noise and the protection of important information,and the complexity of the model is reduced.