Image denoising based on adaptive fractional partial differential equations

Abstract

By combining fractional differential operator which can enhance image texture information with variational partial differential equation and then applying to image denoising, a denoising model based on fractional partial differential operator is put forward. The model can not only better suppress noise of the image, but also better preserve detailed texture information. However, the order of the fractional differential operator must be determined by a lot of experiments. Therefore, calculating local variance is chosen to express the complexity of the local image texture so as to realize adaptive determination of the order of fractional differentiation. The experimental results show that adaptive fractional partial differential operator not only inherits advantages of TV model, but also better keeps detailed image information. The adaptivity of the model is conductive to its being promoted to practical application.