Researches On Denoising Methods In Medical X-ray Digital Image Based On Wavelet Transform

Each process (collection, production, copying, scanning, transmission and transformati- on, etc.) of medical image digital processing will be the inevitable introduction of noise. And yet noise is a crucial factor to the impact of medical image quality. The existence of noise will decrease peak signal noise ratio, making the image become blurred, or even a direct result of the details of some features of images submerged in noise so that the image can not be recognized, especially for the focus,thus affecting physicians'accuracy of diagnosis, so it is necessary to minimize noise impact on the medical image.This paper introduces the wavelet transform theory, describes in detail several wavelet-domain denoising classical methods, and examine the number of medical X-ray film images of the noise distribution, mainly to do the following:1. Sum up the advantages and disadvantages of existing algorithms and carry out the existing denoising algorithms of medical images, in order to provide the basis for better algorithms.2. Based on the intra-layer approach related to the Neigh shrink, put forward a new wavelet domain denoising method. In the contraction threshold processing of the wavelet coefficients, dual-threshold contraction of the wavelet coefficients is used.3. Based on the relevance of the inter-layer theory, the singularity of image, set a new applicability threshold, and build the new soft and hard threshold function.As a result, those coefficients in and near the edge of their relatively small absolute value also may be retained and can remove the details of cause by noise in order to protect the details of medical images.This paper presents two algorithms to overcome the shortcomings of hard threshold and soft threshold function , the denosing results show they are more optimal in the numerical and visual,the algorithm based on the intra-layer is suitable for the abrasion image as well as new, there are good prospects. Experiments show that our new method is effective and available to maintain the detail of the image.