H2TF for Hyperspectral Image Denoising: Where Hierarchical Nonlinear Transform Meets Hierarchical Matrix Factorization

Recently, tensor singular value decomposition (t-SVD) has emerged as a promising tool for hyperspectral image (HSI) processing. In the t-SVD, there are two key building blocks: (i) low-rank enhanced transform and (ii) accompanying characterization of transformed frontal slices. Previous t-SVD methods mainly focus on developments (i), while neglecting other important aspect, i.e., exact this letter, we exploit potentiality in both blocks by leveraging \underline{\bf H}ierarchical nonlinear matrix factorization to establish new T}ensor F}actorization (termed H2TF). Compared shallow counter partners, e.g., or its convex surrogates, H2TF can better capture complex structures slices due hierarchical modeling abilities. We then suggest H2TF-based HSI denoising model develop an alternating direction method multipliers-based algorithm address resultant model. Extensive experiments validate superiority our over state-of-the-art methods.