Statistical Noise Model in Digital Radiographic Images
Jeonghwan Hwang, Jung-Min Hwang, S. K. Park · 2008
In digital radiographic system the noise is originated from various elements of the system, such as camera, X-ray source, imaging screen, inspected object, electronic controller and etc. Radiographs are regarded as being obtained by corrupting a noiseless image. This noise influence image quality seriously and can significantly degrade it. Therefore, before applying denoising techniques, analytic noise model is to be investigated. In this paper, we propose an appropriate approach for the analysis of the noise characteristics in steel-tube radiographic images to verify how noise influence image formation. To achieve a noise model, both global and region-based local analysis is used due to special and statistic estimation of noise. The assumption of a distribution of Gaussian and Poisson distribution causes the robust estimation of the noise variance in regions of high and low intensity. Examples of noise characteristics and its model of steel-tube radiographs show the effectiveness of the proposed approach.