Non-causal Signal Processing Based Fractional Image Denoising Algorithm

With the automatic driving,artificial intelligence,video technology,face recognition,VR technology become popular,image technology of the basic disciplines is also more and more critical field.The images obtained in reality are mostly disturbed by noise.Because of the cause and the diversity of noise,there are often more than one noises in one image which blurs the characteristic information of the image.In order to make the subsequent understanding and analysis of the image more accurate,image denoising is necessary as the image preprocessing work.In recent years,fractional calculus has become more important in basic research and engineering applications,such as digital signal processing,digital image processing and control theory.With the increasing of computer computing speed,fractional calculus become much favored by some experts and scholars.In this paper,we mainly analyze the relationship between forward and backward fractional integral and causal system and anti-causal system in signal processing,analyze their amplitude-frequency characteristics and phase-frequency characteristics,and propose a new Non-causal Signal Processing Based Fractional Low-Pass Filter.The filter has three characteristics: First,in the effective smoothing noise while effectively preserving the edge,texture and other details of the information.Second,according to its ability to suppress the causal filter caused by the phase shift,making the image without phase shift after filtering.Thirdly,the fractional order is determined by combining the value of the texture feature entropy with the image gradient value to achieve the ability to denoise according to the degree of image texture richness.In this paper,the theoretical basis of the algorithm and the principle of filtering are introduced in detail,theoretically the filtering principle is verified,and simulation experiments is maked using a large number of data,the classic image and the actual medical image are compared with the existing image denoising algorithm.Experiments show that the algorithm proposed in this paper has not only improved in subjective vision,but also has better performance in quantitative comparison.