Pan-denoising: Guided Hyperspectral Image Denoising via Weighted Represent Coefficient Total Variation

TL;DR

This work introduces pan-denoising, a paradigm that uses Panchromatic (PAN) images to externally guide hyperspectral image (HSI) denoising. It proposes Panchromatic Weighted Representation Coefficient Total Variation (PWRCTV), a weighted TV regularization on representation coefficients that factors PAN gradient information and slice-wise correlations to preserve textures while denoising; the model is optimized via ADMM. Across synthetic and real PAN-HSI datasets, PWRCTV outperforms state-of-the-art internal-prior methods in denoising quality and also enhances downstream classification performance, with additional validation in hyperspectral pan-sharpening. The approach offers a practical preprocessing step that improves both image fidelity and subsequent analysis, and the authors provide public datasets and code for reproducibility.

Abstract

This paper introduces a novel paradigm for hyperspectral image (HSI) denoising, which is termed \textit{pan-denoising}. In a given scene, panchromatic (PAN) images capture similar structures and textures to HSIs but with less noise. This enables the utilization of PAN images to guide the HSI denoising process. Consequently, pan-denoising, which incorporates an additional prior, has the potential to uncover underlying structures and details beyond the internal information modeling of traditional HSI denoising methods. However, the proper modeling of this additional prior poses a significant challenge. To alleviate this issue, the paper proposes a novel regularization term, Panchromatic Weighted Representation Coefficient Total Variation (PWRCTV). It employs the gradient maps of PAN images to automatically assign different weights of TV regularization for each pixel, resulting in larger weights for smooth areas and smaller weights for edges. This regularization forms the basis of a pan-denoising model, which is solved using the Alternating Direction Method of Multipliers. Extensive experiments on synthetic and real-world datasets demonstrate that PWRCTV outperforms several state-of-the-art methods in terms of metrics and visual quality. Furthermore, an HSI classification experiment confirms that PWRCTV, as a preprocessing method, can enhance the performance of downstream classification tasks. The code and data are available at https://github.com/shuangxu96/PWRCTV.

Figures (12)

• Figure 1: (a) The difference between internal information modeling and external information modeling. Here, $\mathbf{P}\in\mathbb{R}^{M\times N}$ is the PAN image, and $\mathcal{X},\mathcal{Y}\in\mathbb{R}^{M\times N\times B}$ are restored and noisy HSIs, respectively. (b) The illustration of pan-denoising problem.
• Figure 2: The function curves of $y=(1-|x|)^q$ with different values of $q$.
• Figure 3: (a) The PAN and the first three RC images. (b) The local coefficient maps between the gradients of the first three slices of RCs and the gradients of PAN images.
• Figure 4: First row: HSI cubes for the (a) Florence and (b) Milan datasets captured by the PRISMA satellite, and the (c) Beijing and (d) Yulin datasets captured by the XG3 satellite. Second row: The corresponding PAN images.
• Figure 5: Visual inception of HSIs before and after denoising using different algorithms on the Florence dataset (band: 58-35-16) for Case 5.