As the space environment becomes more complex, efficient detection and tracking of resident space objects (RSOs) in Earth's orbit is increasingly important. However, current single-satellite detection systems are limited by observation range and computational power, making real-time RSO awareness difficult. To address this challenge, we propose a collaborative detection model with reconfigurable satellites that leverages a federated learning framework to enable situational awareness and object detection across multiple nodes. Specifically, we designed a Modular Self-Reconfigurable Satellite consisting of four modules, with each module equipped with two cameras. By stitching together their fields of view, the constellation offers comprehensive situational awareness of the surrounding 2π annular space. The approach overcomes the limitation of single-point computing power by aggregating and updating network parameters among reconfigurable satellites, effectively addressing the risks associated with multi-modules data transmission. We conduct extensive experiments to validate the effectiveness of our proposed reconfigurable satellite in detecting RSOs. Experimental results show that the optical system constructed in this paper can effectively perceive 6th magnitude stars with an exposure time of 1 second, and the proposed detection model exhibits a maximum increase in accuracy of 5.2% compared with the single satellite model.

Mirror scanning image mode (MSIM) effectively expands the observation range while maintaining the resolution and has become a prominent research focus in the field of Earth observation; however, it faces the inherent limitation of discontinuous imaging, which poses a major obstacle to achieving seamless multistrip coverage. Additionally, while MSIM expands the imaging range, it also introduces complex time-varying relative motion between the image and focal plane, further increasing the difficulty of image motion compensation. Effectively addressing the limitations of discontinuous imaging and image motion has, therefore, become a critical challenge in the advancement of MSIM. In order to address these difficulties, we introduce a novel image mode, namely deceleration-based MSIM (DMSIM). This mode integrates the satellite’s pitch maneuver with MSIM, effectively expanding the overlap and aiding in image motion compensation. Specifically, DMSIM first constructs a set of parameter constraint equations to define the reasonable range for satellite and payload motion speeds, thereby ensuring seamless coverage of the target area. Subsequently, within the defined speed range, image motion compensation is achieved by designing the satellite’s pitch attitude maneuver; moreover, scaled-down tests and digital simulations are designed to validate our mode. Experiment results indicate that this mode achieves a swath width of$600\times 600$km while the ground sample distance is at the meter level. Meanwhile, the average modulation transfer function (MTF) is 0.088, highlighting its great potential for broad coverage and sharp imaging.

The rotation of camera during scanning imaging induces the image degradation, which means both resident space objects (RSOs) and stars have the same morphological characteristics in the image. It is difficult to detect RSOs from the stellar background in the scanning imaging mode. In this article, we propose an effective motion perception (EMP) approach to detect RSOs by constructing and perceiving morphological differences between moving RSOs and stationary stars in the scanning imaging mode. Specifically, EMP uses the time-delay integration (TDI) technology to compensate for the image degradation caused by the camera’s rotation, so that different morphologies in the image appear between RSOs and stationary stars. A theoretical method of image motion velocity based on the rigorous imaging model and attitude dynamics is proposed to design TDI imaging parameters, which can accurately compensate for the image degradation. Then, dynamic convolutional-oriented bounding box (DC-OBB) based on the adaptive points representation is used to capture the morphological differences between arbitrarily oriented RSOs and stars. Moreover, a linear perception mechanism based on the streak-like characteristics of RSOs is proposed to choose the representative points during training. It makes the detection head precisely classify and locate targets by increasing the sensitivity of deformable aspect ratio. Experimental results show that our method achieves competitive performance on the high-precision-oriented object detection task.

Reference-based super-resolution (Ref-SR) is a heated topic distinguished from single-image super-resolution (SISR). It aims at transferring more texture details from reference (Ref) image with a different perspective, to super-resolve the low-resolution (LR) image. However, the development of Ref-SR within remote sensing (RS) community is limited by three problems. First, RS images exhibit more complex texture details compared to natural images. It’s challenging to learn and reconstruct fine texture of RS images. Second, the lack of high-quality RS image dataset, which contains massive RS image pairs from different perspectives, hampers the model training and diminishes the generalization of Ref-SR within RS community. Third, the lack of physical system prevents it from verifying the feasibility of Ref-SR in RS practice. To address these problems, this paper proposes a novel reference-based gradient-assisted texture-enhancement GAN (RGTGAN), a novel dataset, namely KaggleSRD, and a novel physical simulation system, namely dual-zoom-lens system (DZLS). Specifically, this paper proposes a gradient-assisted texture-enhancement module (GTEM) to fully release the potential of gradient branch to learn fine structures during feature extraction process, a novel dense-intern deformable convolution (DIDConv) to boost the alignment effect between features from different image branches during feature alignment process, and a novel dense-restore-residual (DRR) module to effectively transfer features. Extensive experimental results on both datasets, RRSSRD and KaggleSRD, demonstrate the superiority of the proposed method over state-of-the-art methods. Furthermore, DZLS verifies promising application prospects of the proposed method. Our code and dataset are publicly available at: https://github.com/stdinR/RGTGAN.

Infrared search and track (IRST) systems require reliable detection of small targets in complex backgrounds. Outlier based methods are prone to high false positive rates due to the resemblance of point-like background features to small targets. The difference image-based method is an effective approach for suppressing point-like background interference; however, it has limitations in detecting slow-moving targets. In this letter, a novel sparse trajectory is proposed for moving target detection in IR videos. With a trajectory growing strategy, two kinds of trajectories from difference images, namely short sparse trajectories and long sparse trajectories, are correlated to avoid the slow-moving targets being dismissed. The strategy matches the trajectories based on the sparse trajectory intensity composed of similarity measures and optical flow consistency. Finally, real targets are extracted from candidate trajectories using trajectory filtering. Experimental results show that, in the scene with point-like background features, our method achieves the best detection rate and lowest false alarm compared to state-of-the-art methods.