T-Hy-Demosaicing: Hyperspectral Reconstruction Via Tensor Subspace Representation Under Orthogonal Transformation

Abstract

This paper aims to solve the problem of the hyperspectral imagery (HSI) demosaicing under a novel subsampling hyperspectral sensing strategy. The existing method utilizes the periodic structure of sub-sampling to estimate a fixed subspace in matrix form from the measurement result, which reduces the representation ability of the subspace in iterations and destroys the intrinsic structure of the tensor. To overcome these drawbacks, we propose a tensor-based HSI demosaicing (T-Hydemosaicing) model with tensor subspace representation, which takes the low-tubal-rankness and the nonlocal self-similarity into account. In particular, we suggest a tensor singular value decomposition based on orthogonal transformation (Tran-based t-SVD) to learn the tensor subspace that possesses a more powerful representation ability. In addition, we develop an effective algorithm to solve the proposed non-convex model under the framework of the proximal alternating minimization (PAM) algorithm. Experiments conducted on simulated datasets illustrate that the proposed method outperforms other comparative methods in both visual and quantitative terms.

Hyperspectral Image Denoising via Correntropy-Based Nonconvex Low-Rank Approximation

“…Additionally, the low-rank factorization, which can reduce the computational cost, is another important branch for data recovery [36]- [38]. Inspired by the matrix factorization, tensorbased factorization methods are extended for HSIs denoising, such as t-product [39], [40], Tucker decomposition [41]- [43], CANDECOMP/PARAFAC (CP) decomposition [44], tensor train decomposition [45] and tensor ring decomposition [46]. Besides, the significant correlation among the images acquired across the HSI spectrum allows for the imposition of a spectral low-rank constraint through subspace representation [47]- [49].…”