报告题目：A novel PDE based image restoration: convection-diffusion equation for image denosing and compression

报告人：施因泽教授（台湾中兴大学）

时间：2012年11月8日（星期四）10：30-11：30

地点：理科楼数学系A-304

摘要：In this talk we present a convection-diffusion equation for processing the image denoting, edge preservation and compression. We compare it with a popular nonlinear anisotropic diffusion model which has been widely implemented in image denoting for Gaussian white noise. Here we show that the convection-diffusion model removes noisy effectively, especially for the mixture of salt-and-pepper noise and Gaussian noise. We propose the modified streamline diffusion method for the discretization of the convection-diffusion model to prevent internal layers because of the discontinuities on the coarse grids of image for while using the coarsening algorithm for image compression. The computation cost through the coarsening processing can be reduced dramatically. Numerical experiments have shown the efficiency of our convection-diffusion filtering model for removing the mixture of Gaussian and salt-and-pepper noises, efficiently and reliably in preserving edges quite satisfactorily.

报告人简介：施因泽教授1998年在美国马里兰大学博士毕业，2003年回到台湾国立中兴大学任教至今。施因泽教授在科学计算、偏微分方程数值解，尤其是在医学影像处理方面有着深入的研究工作。

联系人：韩厚德