Xunqiang Gong

1. Building extraction is a fundamental task in remote sensing that underpins urban planning and numerous geoscience applications (Tian et al. 2023). Firstly, it enables urban planning and dynamic ...

Open‐pit mining areas in remote‐sensing imagery present significant challenges for object detection due to complex feature compositions, large scale variations, weak inter‐class discriminability, and sparse spatial distributions. Although single‐stage object detection models provide high computational efficiency and strong real‐time performance, their detection accuracy remains limited when applied to such complex scenarios. To address this issue, this study proposes a lightweight similarity‐aware network with residual multi‐scale bidirectional features, termed LBS‐YOLO, for object detection in open‐pit mining areas. In the neck network, a lightweight residual feature extraction and multi‐scale cross‐kernel perception integration module is introduced to enhance contextual information capture across multiple scales. A bidirectional feature fusion structure is further incorporated in the connection layer to improve multi‐scale feature interaction and better handle objects with diverse spatial scales. In the detection head, a similarity‐aware activation module dynamically adjusts attention weights according to feature statistics, reducing model complexity while enhancing feature representation capability. Experiments conducted on the China's Northern Open‐pit Mining Area dataset and the Brick Kiln dataset demonstrate that LBS‐YOLO achieves mAP _50–95 values of 61.765% and 66.864%, respectively. The results indicate that the proposed method provides accurate and robust detection performance for complex open‐pit mining environments.

As a major economic corridor and ecological barrier in China, the Yangtze River Economic Belt (YREB) has experienced substantial land-use changes that have significantly influenced regional landscape patterns and sustainability. However, the long-term evolution of landscape patterns and their driving mechanisms remain insufficiently understood, particularly considering the spatial heterogeneity among different regions. Based on land-use datasets from 1993 to 2023, this study integrated land-use transfer analysis, grid-based landscape metrics, and the Geodetector model to examine the spatiotemporal evolution of landscape patterns and their driving factors across the YREB. The results showed that cultivated land decreased from 34.63% to 32.78%, forest land increased from 51.51% to 52.36%, and built-up land expanded from 1.36% to 3.61%. Rapid urban expansion was the primary driver of landscape pattern change, leading to increased fragmentation and landscape complexity. Significant regional differences were observed, with relatively stable landscape structures in the upper reaches, transitional changes in the middle reaches, and intensive landscape reorganization in the lower reaches. Both natural and socioeconomic factors influenced landscape evolution, while their interactions generally exerted stronger explanatory power than individual factors. These findings provide valuable support for ecological conservation, land-use optimization, and territorial spatial planning in the YREB.

1. The explosive development of sensors has led to increasingly abundant, accessible remote sensing imagery (Gong et al. 2024; Vivone et al. 2025; M. Zhou et al. 2022). High-resolution multi-spectr...

Pan-sharpening aims to generate high-quality imagery by fusing the spatial details of panchromatic images with the spectral information of multispectral images. However, existing hybrid networks combining the convolutional neural network (CNN) and Transformers often struggle to balance local feature extraction with global context decoding. To address this, we propose a Swin Transformer V2-based multiscale stepwise up-sampling network (SwinMSU). SwinMSU employs a dual-branch architecture: a Swin-based branch captures global contextual representations, while a CNN branch focuses on local spatial features, achieving complementarity via multi-level injection. We introduce the Group Convolution Weight Smooth Adjustment Module, which applies adaptive smoothing regularization to enhance attention on key objects. Additionally, Double/Triple Hierarchical Up-sampling Modules are designed to progressively reconstruct high-resolution features while preserving high-frequency details. Experiments on WorldView-3, GaoFen-2, and QuickBird datasets demonstrate that SwinMSU achieves leading performance (QNR up to 0.9264, 0.9023 and 0.9050), outperforming state-of-the-art methods in both spatial fidelity and spectral consistency. The code can be available at RSIDEA-ECUT/SwinMSU.

Flood change detection from remote sensing data can be used to identify post-disaster flooded areas, providing decision support for emergency rescue and post-disaster reconstruction. Although the combination of SAR and optical images effectively addresses obscuration by clouds and rain, the inherent difference in their imaging mechanisms poses a challenge to improving the accuracy of flood area change detection. Furthermore, existing flood inundation change detection methods based on heterogeneous remote sensing imagery struggle to distinguish small ground objects within the background from the actual inundated regions. Therefore, a pseudo-Siamese hybrid dilation spiral feature network (PSiam-HDSFNet) is proposed in this paper. Firstly, the feature extraction pipeline progressively processes optical and SAR images through five-layer Enhanced Deep Residual Blocks and five-layer Residual Dense Blocks, respectively. A Hybrid Dilated Pyramid (HDP) module based on a sawtooth wave-like dilated coefficient is designed to enhance multi-scale semantics of deep features in order to selectively reinforce semantic features in flood areas and weaken the noise semantics from small ground objects. Then, a Spiral Feature Pyramid (SFP) module is designed to make the deep features of SAR and optical images more consistent in spatial structure and numerical distribution patterns, so that the features of flood areas become more prominent while the noise semantics from small ground objects are further suppressed. After that, the Galerkin-type attention with linear complexity is introduced to the decoder, rapidly reconstructing the abstract semantic information of floods into interpretable flood features. Finally, the Align OPT-SAR (AlignOS) method is designed to align SAR and optical image features, enabling subsequent flood area detection. Seven metrics are adopted in the comparison between PSiam-HDSFNet and the other 14 methods. The results indicate that PSiam-HDSFNet improves change detection accuracy by extracting and processing depth features of these two images without image domain translation, and its F1 scores are improved by 7.704%, 7.664%, 4.353%, and 1.111% in the four flood coverage categories detection tasks compared to the suboptimum.

Optical imagery is often contaminated by clouds to varying degrees, which greatly affects the interpretation and analysis of images. Synthetic Aperture Radar (SAR) possesses the characteristic of penetrating clouds and mist, and a common strategy in SAR-assisted cloud removal involves fusing SAR and optical data and leveraging deep learning networks to reconstruct cloud-free optical imagery. However, these methods do not fully consider the characteristics of the frequency domain when processing feature integration, resulting in blurred edges of the generated cloudless optical images. Therefore, an adaptive frequency domain feature reconstruction-based cloud removal method is proposed to solve the problem. The proposed method comprises four key sequential stages. First, shallow features are extracted by fusing optical and SAR images. Second, a Transformer-based encoder captures multi-scale semantic features. Subsequently, the Frequency Domain Decoupling Module (FDDM) is employed. Utilizing a Dynamic Mask Generation mechanism, it explicitly decomposes features into low-frequency structures and high-frequency details, effectively suppressing cloud interference while preserving surface textures. Finally, robust information interaction is facilitated by the Cross-Frequency Reconstruction Module (CFRM) via transposed cross-attention, ensuring precise fusion and reconstruction. Experimental evaluation on the M3R-CR dataset confirms that the proposed approach achieves the best results on all four evaluated metrics, surpassing the performance of the eight other State-of-the-Art methods. It has demonstrated its effectiveness and advanced capabilities in the task of SAR-optical fusion for cloud removal.