抄録
In this paper, the character of the noise in a medical X-rays image is first analyzed for the purpose of applying to the amount image of low incidence X-rays. Reasonable dominant noise in a medical X-rays image domain is X-rays quantum noise. This X-rays quantum noise is the spatial and time statistical noise according to a Poisson distribution. Second, a poisson distribution degradation image model is examined and a Poisson distribution degradation picture is created. Then, the average filter of the conventional method, a median filter, and a winner filter are compared with Wavelet Shrinkage filter processing to the created Poisson distribution degradation image. When a PSNR value is compared, since the good result was obtained, the way of a Wavelet Shrinkage filter proposes the new method based on a nonlinear Wavelet Shrinkage filter. This technique has the feature that useful information, such as edge of a image, cannot be spoiled but quantum noise can be removed compared with the conventional method. Moreover, as for the proposal technique, it turned out that the PSNR value is improved from the conventional technique.
