Nonlinear diffusion with multiple edginess thresholds

摘要

Nonlinear diffusion has been shown very effective in edge-preserved smoothing of images. However, the existing nonlinear diffusion models suffer several drawbacks: sensitivity to the choice of the conductance parameter, limited range of edge enhancement, and sensitivity to the selection of evolution time. Through analysis we have found that these problems are due to the fact that the conductance parameter is made to be fixed in those nonlinear diffusion models. In this paper, we propose to modify the models in such a way that the conductance parameter is allowed to vary as a function of the diffusion process. It is shown that with proper but not critical choice of this function the above problems can be alleviated. The improvement is illustrated by experiments.