偏微分方程图像平滑模型的一种最优停止准则

基于偏微分方程的各种非线性图像平滑模型, 可归结为求解初始值为输入图像的非线性扩散方程. 这些非线性图像平滑模型在滤除噪声的同时, 在保留图像重要特征方面表现出良好的性能.采用这些模型对图像进行平滑的过程中,迭代停止准则对图像平滑的效果有着重要的影响.该文提出了一种简单实用的最优停止准则,该准则确定了一个标准,使得平滑后的图像与噪声的相关性最小时停止迭代. 它易于实现而且又有很强的通用性.实验结果表明该算法可以在图像平滑过程中适时地停止迭代, 获得满意的图像平滑效果.

An Optimal Stopping Criterion for Image Smoothing Model Based on PDE

A fundamental feature of image smoothing model based on partial differential equation(PDE) is the necessity to decide when to terminate the iteration and the stopping time has a strong effect on filtered image.A simple and effective optimal stopping criterion for image smoothing model based on partial differential equation(PDE) is proposed.This criterion gives a rule for selecting optimal stopping time so that the correlation of filtered image and noise is minimized.This novel method can be used in many kinds of smoothing filtering model based on PDE without needing additive knowledge about image noise.Experimental results have demonstrated that the proposed method could effectively obtain satisfactory filtered images.