基于多尺度残差融合网络的高光谱图像融合算法

Hyperspectral Image Fusion Algorithm Based on Multi-scale Residual Fusion Network

针对高光谱图像空间分辨率低的问题,提出了一种基于多尺度残差融合网络的高光谱图像融合算法。该算法采用多尺度特征提取机制和多级尺度融合机制,将同一目标的低分辨率高光谱图像(LR-HSI)与高分辨率多光谱图像(HR-MSI)融合,得到高分辨率高光谱图像(HR-HSI)。实验证明,论文所提出的方法在CAVE和Harvard这两个数据集上峰值信噪比分别达到了39.8504和42.9646,光谱角映射分别达到了0.0685和0.1585,比其他几种方法均有一定的提升。该算法避免了传统融合方法人为制定融合规则的缺点,在极大程度上降低光谱失真的同时提升高光谱图像的空间分辨率,具有很好的应用前景。

Aiming at the problem of low spatial resolution of hyperspectral images,a hyperspectral image fusion algorithm based on multi-scale residual fusion network is proposed.The algorithm uses multi-scale feature extraction mechanism and multi-scale fusion mechanism to fuse the low resolution hyperspectral image(LR-HSI)and high resolution multispectral image(HR-MSI)of the same target to obtain high resolution hyperspectral image(HR-HSI).Experiments show that the peak sig⁃nal-to-noise ratio of the proposed method in CAVE and Harvard data sets reaches 39.8504 and 43.9646 respectively,and the spec⁃tral angle mapping reaches 0.0685 and 0.1585 respectively,which is better than other methods.This algorithm avoids the shortcom⁃ings of traditional fusion methods that artificially formulate fusion rules,and greatly reduces spectral distortion while improving the spatial resolution of hyperspectral images.It has a good application prospect.