Low-light enhancement is a crucial task in computer vision; it can improve either the subjective experience of viewers or the usability of computer vision systems designed for normal-light images. In this paper, a variational Retinex model in the image domain is developed for low-light enhancement, which infuses classical/fractional differentiation of the input image into the illumination/reflectance component by means of a structure/texture-aware map (SAM/TAM). Firstly, the SAM (TAM) is generated by the inverse square of classical (fractional) differentiation of the input image. Secondly, the regularization term of illumination (reflectance) is defined by utilizing the SAM (TAM) as a weighted matrix, and an illumination guidance term is incorporated into the objective function. The illumination guidance term encourages the estimated illumination to encompass more structural information by penalizing deviation of illumination from the illumination pre-estimated by a dark channel prior to a guided image filtering. Finally, an alternative algorithm is employed to solve the minimization problem involved in the model. The performance of the proposed method is evaluated on three datasets for low-light enhancement and compared with eight state-of-the-art Retinex methods, qualitatively and quantitatively. Evaluation results show that the proposed method generally achieves higher performance in terms of low-light enhancement.

Hyperspectral and multispectral image fusion (HMIF) aims to integrate the high spectral resolution of hyperspectral remote sensing image (HSI) with the high spatial resolution of multispectral remote sensing image (MSI) to reconstruct a high-resolution hyperspectral image (HR-HSI). However, existing deep learning-based methods often show limitations in modeling global dependencies and in jointly optimizing spatial and frequency domain features. To address these challenges, this study proposes a Spatial-Frequency Vision Mamba enhanced Edge-Gaussian cooperative fusion Network (SFVEGNet). Specifically, a VMESS module is developed to capture global contextual information, which is constructed upon a Two-Dimensional Selective State-Space (SS2D) structure and a Gated Feed-Forward Network (GFFN). Furthermore, a FAMF module is designed to achieve complementary feature fusion in the frequency domain. In addition, a Cross-EGCF module is proposed. It utilizes Scharr edge operators and Gaussian smoothing to generate adaptive weights, which enable dynamic cooperative fusion between spatial and frequency domain features. Experiments are conducted on the IEEE 2018, PaviaC, PaviaU, and Chikusei datasets. The experimental results demonstrate that SFVEGNet outperforms existing methods on multiple evaluation metrics. The source code of SFVEGNet will be publicly available.