Vol.38 No.2

Journal of Xi'an Jiaotong University

Feb.2004

retue.gif (1614 ×Ö½Ú)

zwb.gif (1647 ×Ö½Ú)

Improvement of Image Adaptive Nonlinear Filtering
Zhang Yuanlin,Zheng Nanning,Yuan Zejian
(School of Electronics and Information Engineering,Xi'an¡¡Jiaotong¡¡University,Xi'an 710049,China)
Abstract:Based on the analysis of nonlinear diffusion of Peronaª²Malik(PM) equation,a novel nonlinear filtering method is proposed for image smoothing. The idea of edge-preserving regularization is introduced through a boundary intensity function. Comparing with PM equation, the diffusion coefficient is not obtained from the gradient magnitude of the image directly, but the edge information (i.e. boundary intensity function) restored from the gradient magnitude based on the mode of the image. Experiments demonstrate that the proposed approach has better accuracy and reliability than PM for the results of edge restoration.
Keywords:anisotripic diffusion;edge-preserving smoothing;nonlinear filtering