Three-values-weighted Iterative Image Denoising Algorithm Based On D-S Evidence Theory

Digital image technology provides multi-directional technical support for modern society,and facilitates the daily life and scientific research of human society.However,in the process of acquisition and transmission of digital images,it is inevitable to introduce a variety of unexpected noise pollution,damage the content of the images,and hinder people from extracting the information in the images.To solve this problem,image denosing technology has emerged as the times require.In recent decades,it has developed vigorously,and a large number of excellent denoising algorithms have emerged.However,at present,the traditional noise filtering algorithm is more one-sided in identifying and judging noisy pixels,and the complexity of the noise filtering algorithm is polarized seriously.Based on the research status and development trend of noise filtering algorithms,this paper studies the image denosing algorithms under impulse noise pollution,and proposes a three-valued weighted iterative image denosing algorithm based on D-S evidence theory.First,the digital image principle and standard noise models are explained.The multi-attribute performance characteristics of impulse noise are mainly analyzed.The classical image denosing algorithms are used to filter the noise of examples.The advantages and disadvantages of the classical algorithm are analyzed.According to the above analysis conclusions,the key problems of the image denosing algorithm in this paper are determined.Secondly,the D-S evidence theory and the improved interval number theory are introduced to construct the mass function for the dual attributes of impulse noise,and the final criterion is generated using the Dempster's evidence combination method.The effectiveness of the noise detection algorithm is verified by an example.Thirdly,a three-values-weighted method is introduced to perform quadratic weight allocation and interpolation fitting based on the signal pixel distribution of the local window to restore the damaged pixels.A size-variable window iterative update template idea is proposed to make full use of the recovered pixel to solve the problem of full noise area being difficult to recover.At last,the image denoising algorithm is realized by MATLAB software,and the standard test image is tested by denoising simulation experiment,which is compared and analyzed with MDBUTM and other algorithms to evaluate the performance of the algorithm in this paper.According to the results of denosing restoration,this paper analyzes the advantages and disadvantages of the algorithm,and puts forward suggestions and opinions for further algorithm research.