关键词: 图像去噪, 分数阶微分, 全变分模型

Abstract: Combining fractional order differential theory with total variation method, a new image denoising model was proposed, which was based on fractional Partial Differential Equation (PDE). Current Total Variation (TV) model was good at denoising and keeping the characteristics of image edge. This new model effectively inherited these advantages; at the same time, it used the fractional differential amplitude-frequency and retained texture details of smooth region effectively. The simulation results show that on one hand, compared with the existing denoising methods, the new model can not only suppress noise better, but also keep the characteristics of image edge better; especially, it is better than the existing integer order partial differential methods since it can retain more texture details. On the other hand, seen from the comparative experiments of PSNR, the model is more effective and practical on image denoising.

Key words: image denoising, fractional differential, Total Variation (TV) model