A marginal diffusion method to reduce impulse noise in color images

摘要

#$%^&*AU2020100898A420200709.pdf#####Abstract. Images often contain impulse noise which may arise from transmission errors, faulty memory, and so on. In this invention, a decision-based marginal diffusion method is proposed for color images to suppress impulse noise. The handling means is twofold. On the one hand, the suppressing operation is manipulated separately between channels as all components in an image are corrupted independently. On the other, the details of the manipulations are a twostages scheme. In the first stage, a devised detector is separately applied on every channel to identify a component is corrupted or not. In the second stage, a non-linear diffusion is performed to those components that are corrupted. That is, if a component is corrupted, nonlinear diffusion is applied, otherwise, it is left unchanged. This invention not only can effectively suppress the impulse noise in the color image, but also can protect the details of images well.