一种基于形态分量的多聚焦图像融合算法

提出一种基于形态分量思想的多聚焦图像融合算法。该方法首先对源图像迭代分解，将其分解为低频和高频两个分量，并用curvelet变换表示低频分量，然后，对低频分量采用高斯差分算子定义图像点的特征活跃度和融合规则，对高频分量的细节特征度量采用加权梯度差的方法来衡量和融合。仿真实验在四组多聚焦图像中进行，除了与传统的图像融合算法做比较外，还与系数绝对值最大法的融合算法进行比较。实验结果表明:该方法在平均梯度、空间频率、信息熵等指标上优于传统的图像融合方法，同时也优于基于系数绝对值最大法的融合规则。

A Novel Algorithm for Multi-fosus Image Fusion Based on Morphological Component

This paper presents a multi-focus image fusion algorithm based on morphological component.Firstly，source images are decomposed into low-frequent components and high-frequent components by iteration，and curvelet transformation is used for low-frequent components.Secondly，the low-frequent components are fused by applying a rule of feature activity which is defined by using the operator of difference of Gaussian，and the high frequent components are fused by means of a rule of details information which is defined by using weighted gradient.Finally,simulation experiments are conducted through four groups of multi-focus images.The experimental results compared with ones by traditional image fusion algorithm and the algorithm of the max absolute value of component coefficients.The experimental results show that the proposed algorithm outperforms other approaches in terms of average gradient，space frequence and entroy .