Method Research On Image Denoising Of Adaptive Order Fractional Partial Differential Equations

With the continuous development of human society,the technical requirements of multimedia information processing are getting higher and higher.From the images,humans can obtain rich external information,digital image processing technology has become a research hotspot.Under a normal circumstance,images are inevitably interfered by various noises in the process of image dealing,which will reduce the image quality.That not only makes the image become blurred,but also affects the subsequent processing of the image segmentation,image compression,etc.It is a major challenge for researchers to find a fast and effective image denoising algorithm.In this thesis,the adaptive fractional order function and the threshold adaptive diffusion function are introduced,and an image denoising algorithm based on adaptive fractional order calculus is improved by combining theoretical analysis and simulation experiment.The main work and conclusions of this thesis are as follows:1.For Gaussian noise,an image denoising algorithm based on adaptive fractional with improved PM model is proposed.It mainly improves the PM model in three aspects:1)In the PM model,if the edge diffusion function is improperly selected,it will lead to serious impact on the denoising image.Therefore,in this thesis,by analyzing five diffusion functions for partial differential equation,and comparing the experimental simulation data and images,then the best diffusion function is chosen;2)In the traditional denoising algorithm,and a large number of experiments are needed to obtain the value of K.To overcome this problem,a method for getting the adaptive threshold value K is proposed.3)The local variance is used to determine the adaptive fractional order function,in order to reduce the computational complexity,the discrete Fourier transform is used for numerically calculating method in the frequency domain.The experimental results show that the proposed model can not only effectively remove noise,but also preserve the texture and edge information of the image.More importantly,the simulation time is greatly reduced.2.For salt and pepper noise,an image denoising algorithm based on exponential adaptive fractional calculus is proposed.Firstly,the local information entropy and gradient feature are combined to determine the local detail feature of the image,and the adaptive fractional order function is constructed;Secondly,the Otsu image segmentation method based on small probability strategy is used to divide the image into three areas which including noise point area,detail texture area and smooth area.The gray values of the noise point area are replaced by the neighborhood mean value,then the minimum gradient mean of the noise points is calculated,and set it as the threshold T in the exponential adaptive fractional order function.Compared with other algorithms,the algorithm proposed in this thesis can remove noise and retain the edge and texture information of the image,so it has better visual effect.