摘要： 提出一种基于数值微分的图像融合方法。用精确求解离散数据导数的数值微分方法取代传统的邻域差分来求解图像梯度，将求解得到的图像梯度应用到基于偏微分方程的图像融合模型中，改进基于偏微分方程的图像融合数值化方案。将改进前后的方法进行对比实验，结果表明，应用改进方法得到的融合图像的质量指标和图像收敛性较优。

关键词: 图像融合, 数值微分, 偏微分方程, 图像梯度, 数值化方案

Abstract: This paper proposes an image fusion method based on numerical differentiation. The classical calculation for gradient, difference of neighboring pixels is originally replaced by numerical differentiation, an accurate resolution for derivative of discrete data. The obtained image gradients are used in the fusion model based on Partial Differential Equation(PDE) so as to improve the numerical scheme based on PDE. Results of comparison experiments on improved method and former method show that the quality index and image convergence of fused image obtained by new method is improved.

Key words: image fusion, numerical differentiation, Partial Differential Equation(PDE), image gradient, numerical scheme

中图分类号: 

• TP391