Total Nuclear Norms of Gradients for Hyperspectral Image Pansharpening
Total Nuclear Norms of Gradients for Hyperspectral Image Pansharpening
复制标题
DOI：
10.1109/igarss39084.2020.9324412
复制
发表时间：
2020-09
期刊：
影响因子：
--
通讯作者：
Ryota Yuzuriha;Ryuji Kurihara;M. Okuda;Ryo Matsuoka
中科院分区：
文献类型：
--
作者：
Ryota Yuzuriha;Ryuji Kurihara;M. Okuda;Ryo Matsuoka
提出了一种基于正则化函数的高光谱图像泛锐化方法。正则化是基于梯度图像的核范数。与传统的低秩先验不同，我们通过最小化与HSI梯度中旋转平面相关的核规范之和来实现基于梯度的低秩近似。我们的方法明确地同时利用了光谱域和空间域的相关性。我们的方法使用单个正则化函数实现高保真图像泛锐化，而无需显式使用任何稀疏性诱导先验（如10，l1和TV规范）。提出的正则化在一些HSI图像上进行了验证，并与最先进的方法进行了性能比较，以证明其优越的性能。
We introduce a pansharpening method based on a novel regularization function for hyperspectral images (HSI). The regularization is based on the nuclear norms of gradient images. Unlike conventional low-rank priors, we achieve a gradient-based low-rank approximation by minimizing the sum of nuclear-norms associated with rotated planes in the gradient of a HSI. Our method explicitly and simultaneously exploits the correlation in the spectral domain as well as the spatial domain. Our method achieves high-fidelity image pansharpening using a single regularization function without the explicit use of any sparsity-inducing priors such as l0, l1 and TV norms. The proposed regularization is validated on some HSI images with performance comparisons to state-of-the-art methods to demonstrate its superior performance.