一种新的PM模型最优扩散参数估计方法

New optimal diffusion parameter estimate method based on PM model

基于偏微分方程的图像非线性扩散滤波,关键是确定合适的扩散机制与参量。在分析非线性扩散性质的基础上,结合图像的结构特征和视觉特性,提出了一种最优梯度阈值和最佳扩散时间尺度的估计方法。实验结果表明该方法与最小均方误差准则、信噪比准则和相关系数最小准则相比,具有更好的稳定性和视觉特性。

The key of image nonlinear diffusion filtering based on PDE（Partial Differential Equations） is to determine the appropriate diffusion mechanism and parameters.Combined with the images＇ structures and visual features, this paper proposes an optimal gradient threshold and optimal diffusion time scale estimate method on a basis of analyzing the characteristics c nonlinear diffusion.The experimental result shows that the new method has better stability and visual features compared wil MSE, SNR and minimum correlation coefficient.