Research On Image Denoising Algorithm Based On Direction Feature

As an effective way of information dissemination,images are very important in today's society.The image will often be disturbed by noise and the noise will directly affect people's understanding of the image.Therefore,how to effectively denoise has been a hot topic of research,This article takes the image direction features as the main line of research and has achieved the following results:(1)A non-local mean filtering denoising algorithm based on edge direction is proposed.First,the Canny operator is used to obtain the image edge matrix;the pixels in the image are divided into edge pixels and non-edge pixels according to the edge matrix;edge pixels search for similar blocks along the edge direction,and non-edge pixels search for similar blocks along the horizontal and vertical directions;The weight of the edge pixels needs to be calculated separately in the noise image and the edge matrix.The weight of the pixels at the non-edge position is consistent with the weight calculation of the traditional non-local mean filtering denoising method.The algorithm has obtained good denoising results,the image is clean and the edges are clear.(2)A frequency domain denoising algorithm based on edge direction features is proposed.In order to make full use of the directional features of the image,we combined the non-local mean filtering algorithm based on the edge direction with the multi-directional non-downsampling shearlet transform to perform image denoising.First,the Canny operator is used to obtain the image edge matrix;through the non-downsampling shearlet transform,the three subbands of low frequency,bandpass,and high frequency of the image are obtained;the coefficients at the edges of the low frequency and high frequency subbands are non-local mean filter based on the edge direction Algorithm processing,the coefficients at the non-edges are processed by traditional non-local mean filtering,and the band-pass coefficients are processed by hard thresholds;the three subbands processed are combined for inverse transformation to obtain the denoised image.The algorithm utilizes the non-downsampling shearlet transform and the directional characteristics of the image edge,and uses different processing methods for the edge pixels and non-edge pixels of different subbands of the image.The algorithm better retains the edge information and eases the vibration of the denoising algorithm.Bell effect,denoising effect has been significantly improved.