Application Study On Digital Filter In Medical Image Denosing Processing

This paper studies some important filtering and denosing technologiesin digital image processing and applies them to medical image denosing processing. The total research direction is divded into two: nonlinear spatial filtering and wavelet transform filtering. About spatial filtering, the most valuable work is the two kinds of improved algorithms about classical Perona filter, which preserve the image edges more effectively. We also use AOS(addictive operator splitting) scheme in anisotropic diffusion equation to improve the competence of edges preservation and achieve region smooth effect. In other nonlinear filtering methods, some improved thinking to median filter and Lee filter are also presented. About wavelet transform filtering, combining with the theory of Lipschipz exponent, we present a unified form of wavelet coefficient threshold and in this section we reveal the relation between wavelet transform based on haar function and Perona filter. In addition, produces of medical images and causes of noise, example for ultrasound Image,magnetic Resonance Image and X-ray image,are also refered to.This article value theory and practice. To every algorithm, there are both theory analysis and relative programs, which are designed by c++ program language or matlab to verify the properties of these algorithms. For the purpose of drawing the conclusion effectively, we especially design a chapter to compare and evaluate the denosing effect of different algorithms through these experiment data and relative criterions. The results of experiments show all improved algorithms have more better performance than unimproved and they fit the needs of medical image denosing. In total, anisotropic diffusion filtering have the best performance. All the methods related in this paper enhance application of digital filter techonology and have theoretic and practical meaning.