The problem of inadequate object detection accuracy in complex remote sensing scenarios has been identified as a primary concern. Traditional YOLO-series algorithms encounter challenges such as poor robustness in small object detection and significant interference from complex backgrounds. In this paper, a multi-scale feature fusion framework based on an improved version of YOLOv8_L is proposed. The combination of a graph attention network (GAT) and Dilated Encoder network significantly improves the algorithm detection and recognition performance for space remote sensing objects. It mainly includes abandoning the original Feature Pyramid Network (FPN) structure, proposing an adaptive fusion strategy based on multi-level features of backbone network, enhancing the expression ability of multi-scale objects through upsampling and feature stacking, and reconstructing the FPN. The local features extracted by convolutional neural networks are mapped to graph-structured data, and the nodal attention mechanism of GAT is used to capture the global topological association of space objects, which makes up for the deficiency of the convolutional operation in weight allocation and realizes GAT integration. The Dilated Encoder network is introduced to cover different-scale targets by differentiating receptive fields, and the feature weight allocation is optimized by combining it with a Convolutional Block Attention Module (CBAM). According to the characteristics of space missions, an annotated dataset containing 8000 satellite and space station images is constructed, covering a variety of lighting, attitude and scale scenes, and providing benchmark support for model training and verification. Experimental results on the space object dataset reveal that the enhanced algorithm achieves a mean average precision (mAP) of 97.2%, representing a 2.1% improvement over the original YOLOv8_L. Comparative experiments with six other models demonstrate that the proposed algorithm outperforms its counterparts. Ablation studies further validate the synergistic effect between the graph attention network (GAT) and the Dilated Encoder. The results indicate that the model maintains a high detection accuracy under challenging conditions, including strong light interference, multi-scale variations, and low-light environments.

The wide-field imaging system carried on a high-altitude or near-space vehicle takes high-resolution images of the ground to measure and map targets. With the improvement of imaging resolution and measurement accuracy, the focal length of the wide-field imaging system is getting longer. The requirement for image motion compensation (IMC) accuracy is getting higher, and the influence of optical path coupling is increasing within the process of two-dimensional (2D) IMC. To further improve the IMC accuracy of the wide-field imaging system, an innovative IMC method is first proposed in this paper. The method is based on the 2D motion of the scanning platform and secondary mirror. Secondly, to solve the optical coupling problem in the process of 2D IMC, the coupling phenomenon is analyzed. The coupling relationships between 2D scanning motion, 2D secondary mirror motion and image motion is derived from the compensation process. A complete 2D IMC model is established, and a 2D IMC method, including an optical path decoupling correct regulator (ODCR), is designed. Finally, the method is verified in laboratory and field flight tests. The results show that the proposed method can effectively correct the coupling error of the optical path in the process of IMC and achieve high-resolution 2D IMC. When the scanning speed is 60°/s and the exposure time is 2 ms, the accuracy of the 2D IMC is up to 0.57pixels (RMS) in the pitch direction, and 0.46 pixels (RMS) in the roll direction.

Spacecraft component recognition is crucial for tasks such as on-orbit maintenance and space docking, aiming to identify and categorize different parts of a spacecraft. Semantic segmentation, known for its excellence in instance-level recognition, precise boundary delineation, and enhancement of automation capabilities, is well-suited for this task. However, applying existing semantic segmentation methods to spacecraft component recognition still encounters issues with false detections, missed detections and unclear boundaries of spacecraft components. In order to address these issues, we propose a Multi-scale Adaptively Spatial Feature Fusion Network (MASFFN) for spacecraft component recognition. The MASFFN comprises a Spatial Attention-aware Encoder (SAE) and a Multi-scale Adaptively Spatial Feature Fusion-based Decoder (Multi-ASFFD). First, the Spatial Attention-aware Feature Fusion Module (SA-FFM) within the SAE integrates spatial attention-aware features, mid-level semantic features, and input features to enhance the extraction of component characteristics, thus improving the accuracy in capturing size, shape, and texture information. Second, the Multiscale Adaptively Spatial Feature Fusion Module (M-ASFFM) within the Multi-ASFFD cascades four Adaptively Spatial Feature Fusion (ASFF) blocks to fuse low-level, middle-level and high-level features at various scales to enrich the semantic information for different spacecraft components. Finally, a compound loss function comprising the cross-entropy and boundary losses is presented to guide the MASFFN better focus on the unclear component edge. The proposed method has been validated on the UESD and URSO datasets, and the experimental results demonstrate the superiority of MASFFN over existing spacecraft component recognition methods.

Pose measurement of non-cooperative targets in space is one of the key technologies in space missions. However, most existing methods simulate well-lit environments and do not consider the degradation of algorithms in low-light conditions. Additionally, due to the limited computing capabilities of space platforms, there is a higher demand for real-time processing of algorithms. This paper proposes a real-time pose measurement method based on binocular vision that is suitable for low-light environments. Firstly, the traditional point feature extraction algorithm is adaptively improved based on lighting conditions, greatly reducing the impact of lighting on the effectiveness of feature point extraction. By combining point feature matching with epipolar constraints, the matching range of feature points is narrowed down to the epipolar line, significantly improving the matching speed and accuracy. Secondly, utilizing the structural information of the spacecraft, line features are introduced and processed in parallel with point features, greatly enhancing the accuracy of pose measurement results. Finally, an adaptive weighted multi-feature pose fusion method based on lighting conditions is introduced to obtain the optimal pose estimation results. Simulation and physical experiment results demonstrate that this method can obtain high-precision target pose information in a real-time and stable manner, both in well-lit and low-light environments.

Accurate in-orbit optical measurement of non-cooperative space objects is hindered by radiometric instability under extreme and spatially non-uniform illumination. Photon-limited acquisition, ultra-high dynamic range, and abrupt shadow transitions disrupt the radiance-to-response mapping in the imaging chain, violating the implicit illumination-consistency assumption in many learning-based perception pipelines. The resulting illumination-reflectance entanglement increases the sensitivity of geometric inference to radiometric perturbations, degrading localization accuracy, increasing measurement uncertainty, and reducing repeatability. Rather than enhancing images for visual appearance, this work addresses illumination-induced degradation as a feature reliability problem in optical metrology. We propose a Physics-Inspired illumination-decoupled perception framework, PIDP-YOLO, that improves illumination-robust feature consistency at the representation level. A Retinex-inspired prior is embedded into multi-scale latent features to decompose them into illumination and reflectance components, separating irradiance attenuation from intrinsic geometric structures. A lightweight Physical Rectification Feature Fusion module compensates radiometric attenuation before semantic aggregation, while radiometric-geometric consistency constraints stabilize end-to-end optimization. To evaluate localization robustness, we construct a dedicated benchmark, SMOD-Dark, under controlled laboratory illumination to assess geometric performance under representative severe radiometric degradation. Under the object-disjoint SMOD-Dark protocol, PIDP-YOLO achieves mAP50 = 0.931 and mAP50:95 = 0.508, outperforming YOLOv13n by 4.1 and 3.1 percentage points, respectively. Measurement uncertainty analysis further shows enhanced boundary reconstruction fidelity and a 24.2% reduction in low-light center-position standard deviation, decreasing from 0.091 to 0.069, while maintaining real-time performance at 138 FPS. These results indicate that the proposed Retinex-inspired feature rectification improves bounding-box-level localization repeatability and measurement consistency under severe illumination variations.

Based on vector wave aberration theory, this paper analyzes the relative positions of the third-order coma node and astigmatism nodes under pupil decenter and proposes an initial structure selection criterion in which the coma node coincides with the geometric midpoint of the two astigmatism nodes. Using this criterion, an F/6 off-axis catadioptric telephoto optical system with a focal length of 900 mm and an entrance pupil diameter of 150 mm was designed. The measured on-axis RMS wavefront error was better than 0.025λ at 632.8 nm. The results demonstrate that the system meets the requirements for high-resolution long-focal-length imaging.