A time-dependent anisotropic diffusion image smoothing method

摘要

In this paper, a time-dependent anisotropic diffusion image smoothing method is proposed with attempting to address the limitations in the traditional algorithms. To this end, we suggest a new diffusion coefficient and set the parameters that are the Gaussian scale and the gradient threshold gradually decreasing with time, which is significant to preserve edge and boundary features. The stopping time is dependent on an iterative SNR measure, so as to avoid the excessive smoothing problem. An efficient numerical schema is used for the method implementation. Experimental results represent that the time-dependent anisotropic diffusion image smoothing method is capable of reducing noise efficiently and preserving shaper boundaries.