基于估计理论的红外与可见光图像融合算法

针对传统图像融合方法鲁棒性差的缺点,设计了一种基于估计理论期望值最大的图像融合方法。从多感测器成像模型出发,设计出图像的形成模型,分别基于图像的模型以及期望值最大方法,推导出估计真实场景的迭代过程,最终得到融合图像。探讨了基于拉普拉斯金字塔、梯度金字塔、小波、小波框架以及方向滤波器金字塔等方法的多尺度展开的融合情况,并与EM方法做了比较,数字仿真表明,图像受到噪声污染时采用EM算法进行融合具有较强的鲁棒性。

An Image Fusion Algorithm with Infrared and Visible Light Based on Estimation Theory

A novel image fusion algorithm is proposed based on EM algorithm.This method infers the iterative process of real scene to fusion images according to EM algorithm based on multi-image model.Varieties of methods,such as Laplace Pyramids,wavelet and gradient pyramids,are discussed and compared with EM algorithm.The digital imitation results show that the robustness of the method is quite good even when fusing the seriously polluted image.