一种新的图像融合及性能评价方法

Novel image fusion algorithm with novel performance evaluation method

通过重新定义交叉熵,并将其与熵相结合,提出了一种新的评价图像融合性能的目标函数,即综合熵。将取亮度最大值和对两幅图像进行加权平均的两类图像融合方法相结合,对上述目标函数进行迭代寻优,给出了一种新的图像融合算法。对包括红外与可见光图像,雷达与可见光图像,以及不同性能的可见光图像等多组图像进行实验,结果表明图像融合新算法和性能评价指标都是有效的。

By revising the definition of the cross entropy and combining it with entropy, a novel image fusion performance evaluation function based on entropy and cross entropy, i. e. , the associated entropy is presented. By use of the associated entropy as the objective function, a novel image fusion algorithm is proposed, which is manipulated by operating the maximum method and the weighted average method alternatively and iteratively, Experiments are done to multiple groups of images, including the fusion of inferred image with optical image, and the fusion of SAR with optical image, and the fusion of optical images with different properties etc. The results concluded from the experiments verify the feasibility and effectiveness of the novel image fusion method as well as the novel performance evaluation function.