Existing image fusion methods have developed increasingly sophisticated network architectures for exploiting modality-shared and modality-specific features. However, despite these advancements in feature extraction, most methods ultimately rely on relatively simple implicit or explicit fusion strategies, which can compromise interpretability and limit fusion accuracy. In this paper, we incorporate visual autoregressive modeling to bridge the gap between implicit feature extraction and explicit modality fusion. First, the proposed approach conducts a low-to-high resolution autoregressive objective with modality-specific features, introducing a scalable feature autoregressive mechanism. It aggregates local and global contextual dependencies while enhancing implicit cross-scale interaction. Furthermore, to promote the consistency and complementarity across modalities, we embed an explicit high-order fusion strategy within the progressive modality-specific feature extraction process. This integration facilitates a next-scale synergistic relationship between implicit learning and explicit fusion. Our High-order Feature AutoRegressive Fusion framework (HFARFusion) provides a robust and interpretable solution for general image fusion tasks, effectively balancing fusion performance and transparency through the strengths of autoregressive learning. Extensive experiments demonstrate the outstanding performance of the proposed method in several classical fusion tasks, including infrared-visible, medical, multi-focus, and multi-exposure image fusion. Our code is available at https://github.com/happysbn/HFARFusion.

Accurate segmentation of water bodies from aerial images is critical for advancing our understanding of climate change, improving flood prevention and mitigation efforts, and supporting ecological monitoring. Recently, deep learning-based methods have made remarkable progress in water body segmentation. However, there still exist a series of challenges in practical applications, including mis-segmentation of low-contrast regions, difficult delineation of complex terrain boundaries, and loss of small water features. While most of existing methods are designed for supervised learning with paired samples, they may also benefit from self-supervised learning techniques such as masked auto-encoders (MAE). In this work, we focus on the water segmentation problem and propose a new water segmentation framework, named MHNet. MHNet integrates a Hybrid Multi-Scale Encoder-Decoder Network, combining convolutional and transformer-based components to effectively capture global context while minimizing computational costs. A key innovation is the adaptation of the MAE mechanism, where masks are applied to multi-scale Restormer block outputs during the training phase, enabling the model to better integrate local and global information, thereby enhancing boundary segmentation accuracy. Additionally, we propose a Multi-Channel Feature Fusion (MCFF) module that synthesizes masked feature maps across scales, reducing redundancy and improving generalization by capturing both fine details and contextual information. Extensive experiments on multiple public datasets demonstrate that MHNet outperforms state-of-the-art methods, highlighting its effectiveness and robustness in water body extraction tasks. MHNet is deployed and performs online predictions via Google’s Vertex AI platform, thereby integrating it into a Geographic Information System (GIS) using Google Earth Engine (GEE) for accurate and efficient extraction of lakes on the Tibetan Plateau.

Creating a comprehensively representative image while maintaining the merits of various modalities is a key focus of current Multi-Modality Image Fusion research. Existing unified methods often struggle to handle varying types of degradation while extracting modality-shared and modality-specific information from source images, leading to limitations in their generative or representation capabilities under different conditions. To address the challenge, we propose MVFusion, a novel self-supervised masked variational autoencoder framework that simultaneously enhances generative training and representation learning. It is designed to cope with varying image quality and dataset composition with a unified framework while ensuring effective fusion of modality information. Specifically, MVFusion employs a self-supervised masked autoencoder to reduce the impact of redundancy and degradation in the source images, and thus learns the latent distribution of degraded input images in the generative training stage. In addition, we incorporate variational feature learning to further preserve the distinctive modality features in the representation learning stage. Extensive experiments demonstrate that our model achieves promising results in several classical fusion tasks, including infrared-visible, multi-focus, multi-exposure, and medical image fusion. The code is available at https://github.com/shiboneng/MVFusion.