摘要
协方差交叉算法是分布式信息融合中不需要计算局部估计误差之间的相关性、通过优化一定的目标函数得到的一种保守的分布式融合估计方法。这种方法为图像融合增强提供了一种新思路。介绍了一维协方差交叉算法,把此方法扩展到二维信号和图像融合上,提出了一种基于协方差交叉算法的图像融合方法,最后对融合后的图像与已有的融合方法进行比较。结果表明,融合效果优于小波方法、经验模式分解方法和非负矩阵分解方法。
Covariance intersection algorithm is a conservative distributed fusion estimation method obtained by optimizing certain objective function, without needing to compute the correlation among local estimation errors, which provids a new idea for image fusion enhancement. Firstly, the 1D eovarianee intersection algorithm was introduced;then this method was extended to 2D signal and image fusion for the first time, and an image fusion method based on covariance intersection was put forward. Finally, the fused image was compared with the existing fusion methods, and the results showed that fusion effect of it was superior to that of wavelet method, empirical mode decomposition method, and non-negative matrix faetorization method.
出处
《电光与控制》
北大核心
2013年第6期4-6,11,共4页
Electronics Optics & Control
基金
国家自然科学基金(60975016
61002052)
浙江大学CAD&CG国家重点实验室开放课题(A1214)
海军大连舰艇学院科研发展基金
关键词
遥感图像
图像融合
图像增强
协方差交叉算法
remote sensing image
image fusion
image enhancement
covariance intersection algorithm
