Hyperspectral image denoising and antialiasing based on tensor space and reciprocal cell

Conventtrial denoising and antialiasing algorithms are usually for single band images. Previously, numerous studies have only designed for single band images. Aiming at the data characteristics of hyperspectral image and the influence of noise and aliasing on the image, a multidimensional filtering algorithm combining tensor and reciprocating cells was proposed and applied to denoising and antialiasing of hyperspectral images. The method introduced the tensor, and the hyperspectral image data was regarded as the third-order tensor expression. The reciprocal cell was used to obtain the spectrum extrapolation which containd less image aliasing and noise. From the point of view of the minimum mean square error, the algorithm alternately iterated to solve the three directions of the filter, and finally completed the image filtering. The algorithm could effectively reduce the image aliasing and noise under the premise of ensuring the consistency of image space and spectral information. The effectiveness of the proposed algorithm was proved by comparing with multiple sets of hyperspectral data of the two-dimensional Wiener filter algorithm and tensor multidimensional denoising algorithm.