Abstract
Hyperspectral image (HSI) restoration is an important task of hyperspectral imagery processing, which aims to improve the performance of the subsequent HSI interpretation and applications. Considering HSI is always influenced by multiple factors—such as Gaussian noise, stripes, dead pixels, etc.—we propose an HSI-oriented probabilistic low-rank restoration method to address this problem. Specifically, we treat the expected clean HSI as a low-rank matrix. We assume the distribution of complex noise obeys a mixture of Gaussian distributions. Then, the HSI restoration problem is casted into solving the clean HSI from its counterpart with complex noise. In addition, considering the rank number need to be assigned manually for existing low-rank based HSI restoration method, we propose to automatically determine the rank number of the low-rank matrix by taking advantage of hyperspectral unmixing. Experimental results demonstrate HSI image can be well restored with the proposed method.
Highlights
Hyperspectral imagery (HSI) is 3D data containing both spatial and spectral information, which is widely used for lots of remote sensing related applications [1,2,3,4,5,6,7]
When we apply HSI restoration methods on real noisy data, we can only evaluate the performance qualitatively since the noise-free images are always absent in reality, i.e., we evaluate the restoration results by the visual effect
It guarantees the proposed method is more suitable for HSI restoration and obtains better restoration performance, which can be seen from Figures 3–11 and Table 1
Summary
Hyperspectral imagery (HSI) is 3D data containing both spatial and spectral information, which is widely used for lots of remote sensing related applications [1,2,3,4,5,6,7]. HSI is unavoidably influenced by multiple kinds of factors—including noise, stripe corruption, etc.—during imaging and acquisition [8], which decrease the image quality. It increases the difficulty for people interpretation or machine understanding [9,10,11,12], which makes HSI restoration an essential step for HSI processing and analysis. As indicated by the name, signal-independent-noise method assumes the generation of noise is independent of signals, while the generation of noise is related with signals for signal-dependent-noise method. These two kinds of methods are proposed for different
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have