Fractional Order Regularization Image Denoising Model Based On Adaptive Order

Image denoising technology is one of the common means used in the field of image processing.Its main function is to improve the clarity of the image by removing the excess noise,so as to meet the visual needs of human.Image denoising method is widely used in the field of science and technology,such as medical image,remote sensing image and so on.Image denoising technology is the process of estimating the clear image based on given noisy image.At present,there are lots of image denoising models,but the image denoising problem is still a hot research field because of the different image features.In the field of image denoising,the classical image denoisng model is the total variation(TV)model,but this model is prone to produce the so-called staircase effect when removing noise.To address this effect,domestic and foreign scholars proposed fractional total variation regularization(FTV)model.FTV models can retain image texture details very well,but its order is difficult to determine.Based on this,the fractional order regularization is studied in detail,and two kinds of adaptive fractional order regularization models are proposed in this paper.The main research contents are as follows:1.An adaptive fractional image denoising(AFTV)model based on image complexity is proposed.The model starts from three aspects of the image,namely the appearance of the image gray level,the spatial distribution of gray level and the appearance of the target object,and the model describes the complexity of the image is described by five factors including information entropy,energy,contrast,correlation and edge ratio.Thus,a fractional order regularization image denoising model based on adaptive order is formed,and then the automatic selection of order is realized.In solving the model,we use the alternating direction method of multiplier(ADMM).Experimental simulation shows that the model is effective in determining the adaptive order.2.An image denoising(AFAD)model based on adaptive fractional anisotropic diffusion is proposed.The model calculates the gradient,information entropy and variance of the image,and combines them linearly to determine the order of the model.In solving the model,we use Fourier transform and gradient descent method.The experimental results show that this model can not only solve the problem of non-adaptive order in fractional regularization model,but also effectively suppress the noise in the image,and accurately capture the texture structure and other fine features of the image.