Remaining Useful Life (RUL) prediction is crucial for implementing predictive maintenance strategies, however, RUL prediction is severely constrained by the lack of high-quality labeled life-cycle data. Moreover, complex coupling relationships exist within the obtained multidimensional degradation data, making it difficult to construct an accurate health index (HI) for the system. To address this challenge, we propose an RUL prediction method based on sequential healthy index evaluation which incorporate two parts: the parameter prediction process and the health index fusion process. The core innovation of this study is an RUL prediction method that integrates a CNN-Transformer hybrid model with a sequential health index evaluation scheme. Compared to traditional data-driven methods, our approach incorporates a chunk-interaction mechanism into the multi-head attention design, thereby reducing model complexity and computational demands. Simultaneously, the sequential evaluation scheme dynamically constructs the health index based on the Mahalanobis distance and the Sequential Evaluation Ratio (SER), which eliminates the reliance on high-quality labeled life-cycle data. Experimental results demonstrate that the proposed method outperforms existing deep learning approaches (such as LSTM, Transformer, and Att-BiGRU) across multiple datasets, exhibiting higher prediction accuracy and robustness, particularly in label-scarce scenarios.

Small object detection in UAV aerial imagery poses significant challenges due to sparse pixel representation and ambiguous object boundaries. Through systematic analysis, we identify three critical degradation stages during forward propagation in deep detection networks: edge attenuation during feature extraction, semantic conflict during feature fusion, and detail loss during feature reconstruction. Existing methods address these stages in isolation or implicitly, lacking collaborative and stage-aware repair strategies. To address this issue, we propose EAGLE-DET, a novel detection framework based on sparse multi-scale attention and refined transformation. Specifically, the framework comprises three core modules: (1) the Cross-stage Multi-resolution Edge Enhancement Network (CMENet), which preserves small object edge representations via adaptive high-low frequency decomposition; (2) the Attention-guided Multi-scale Feature Fusion Network (AMFFN), which resolves cross-scale semantic conflicts through pyramidal sparse attention and multi-scale spatial decoupling; (3) the Enhanced Upsampling with Channel Bridging and Spatial Coordination module (EUCBSC), which recovers spatial detail fidelity via bidirectional channel shift mixing. Extensive experiments on three benchmark datasets-VisDrone-2019, UAVDT, and DOTA1.0-demonstrate the effectiveness of EAGLE-DET, which achieves improvements of 4.5% AP50 and 2.9% AP50:95 on VisDrone-2019 over the baseline, while maintaining inference at 71.7 FPS, achieving an optimal accuracy-efficiency trade-off.

Real-time object detection networks based on UAV have been used in various fields. However, some challenges need to be solved: (1) Conventional detection algorithms are not suitable for small targets; (2) The computational capacity of the UAV platform is limited; (3) The sample distribution in the aerial dataset shows the characteristics of long-tail distribution. Categories at the tail end often need to be better learned. To address these challenges, we propose the AIR-YOLO-pruned method, a lightweight UAV-based object detection method built on the YOLOv8. In this paper, we propose the AIR-YOLO which is suitable for small object detection. We introduce the gradient adaptive allocation loss to enhance the model's learning ability for tail categories. To eliminate redundant components in AIR-YOLO, we design a kind of structured pruning strategy. Experiment results indicate that our AIR-YOLOn-pruned method, with competitive computational cost, achieves a 17% improvement in accuracy compared to YOLOv8n.

Current cross-modal feature fusion research mainly adopts the deep features of the last layer of the backbone network as inputs, ignoring the utilization of detailed information in the shallow features of the backbone network, leading to certain limitations of the model in coping with the various challenges of rapid changes of the target in cross-modal images. To solve this problem, this paper proposes a novel tracker based on the Multiscale State Space Attention Fusion Network (MCSSAFNet), which realizes the learning and fusion of different modal feature information at different scales by introducing Mamba. On this basis, an adaptive-aware loss function is proposed to adaptively weight the classification loss firstly, to solve the imbalance between the classification score and the localization score by enhancing the learning attention to the difficult samples, and to improve the ability to discriminate the difficult targets. Adaptive weighting is then performed for IoU loss to enhance the learning of high-quality samples while improving the learning of low-quality samples, which in turn improves the model IoU accuracy. Comprehensive experimental validation is carried out on four mainstream RGBT open tracking datasets, namely, RGBT210, RGBT234, LasHeR, and VTUAV, and the experimental results show that the tracking performance of the proposed algorithm outperforms the existing algorithms and achieves a running speed of 37 fps on a GTX 3090 GPU.

This paper focuses on dual-angle constraint of the impact angle and field of view (FOV) angle against maneuvering target. Firstly, the collision triangle theory is used to transform the guidance model constrained by the impact angle into a nonlinear second-order model about the satisfaction of final light-of-sight (LOS) angle by associating LOS angle with impact angle. Secondly, a guidance law with desired impact angle is proposed based on the state feedback method, and a nonlinear state observer is established to estimate the target acceleration term, which is unknown in engineering practice. Moreover, the design of the bias term instruction is completed in accordance with the switching logic method in order to realize the FOV constraints. It is worth mentioning that the selection of the middle point effectively alleviates the problem that the guidance command may jump violently during switching. Simulations presented to guarantee the effectiveness and robustness of proposed guidance strategy.

This paper introduces a lightweight remote sensing image dehazing network called multi-dimensional weight regulation network(MDWR-Net),which addresses the high computational cost of existing methods.Previous works,often based on the encoder-decoder structure and utilizing multiple upsampling and downsampling layers,are computationally expensive.To improve effi-ciency,the paper proposes two modules:the efficient spatial resolution recovery module(ESRR)for upsampling and the efficient depth information augmentation module(EDIA)for downsampling.These modules not only reduce model complexity but also enhance performance.Additionally,the partial feature weight learning module(PFWL)is introduced to reduce the computational burden by applying weight learning across partial dimensions,rather than using full-channel convolution.To overcome the limitations of convolutional neural networks(CNN)-based networks,the haze dis-tribution index transformer(HDIT)is integrated into the decoder.We also propose the physical-based non-adjacent feature fusion module(PNFF),which leverages the atmospheric scattering model to improve generalization of our MDWR-Net.The MDWR-Net achieves superior dehazing performance with a computational cost of just 2.98×109 multiply-accumulate operations(MACs),which is less than one-tenth of previous methods.Experimental results validate its effectiveness in balancing performance and computational efficiency.

In response to the multiple challenges faced in detecting small infrared moving targets—such as cluttered backgrounds, limited object size, weak feature representation, and low detection precision—this study introduces an enhanced detection model tailored for UAV infrared imagery. The proposed approach, named IFMO-YOLOv11, is built upon the YOLOv11 architecture and specifically optimized to improve the recognition of small infrared objects in aerial scenarios. Firstly, this model utilizes dilated convolution to design the multi-layer feature dilated convolution module (MLF-DC), which replaces the original SPPF layer to enhance the extraction of detailed features in UAV images. Secondly, to strengthen the C3K2 structure, the RFCBAMConv module is incorporated, refining internal convolutional mechanisms and feature integration, thereby boosting the model’s feature extraction performance. In addition, the integration of the Biformer module allows the network to better attend to critical details of small targets. To further enhance localization precision, an improved bounding box regression loss function—power-WIoU (P-WIoU), based on the upgraded Wise-IoU v3—is introduced for more accurate prediction box positioning. Through rigorous evaluation and testing on a publicly available dataset with wide recognition, extensive experimental results demonstrate that the proposed model achieves superior overall performance compared to other popular approaches. Notably, it offers a marked enhancement in detecting small, moving infrared targets.

Image matching faces challenges in maintaining robustness under complex conditions such as sparse textural regions, scenes with repetitive structural patterns, and large viewpoint variations, where neither point-based nor line-based feature matching alone can achieve consistent performance across all scenarios.This paper presents a novel graph neural network-based point-line joint matching algorithm designed to overcome these limitations. First, we introduce a connection matrix representation that unifies point and line features within a graph structure, eliminating the need for expensive line descriptor extraction while preserving geometric relationships. Second, we design a dynamic graph neural network incorporating self-attention, linear propagation, and cross-attention mechanisms to capture both intra-image and inter-image feature associations, enhancing global consistency and discriminative power. Third, we employ a Dual-Softmax matching strategy combined with negative log-likelihood loss to achieve efficient and robust matching even in repetitive structural scenarios. Extensive evaluations on the Wireframe dataset confirm our method’s superiority, achieving 35.3% AUC-RANSAC@1px for point matching and 0.60 mAP for line matching, while maintaining real-time efficiency.Finally, we evaluate our method on the task of aircraft pose estimation.

Infrared and visible image (IVI) fusion retains the advantages of both modalities and is widely used in advanced vision tasks such as object detection. To improve the quality and visualization of fused images, a dual autoencoder (dual-AE) and latent-guided diffusion fusion network (D2Fusion) for IVI is proposed. Firstly, D2Fusion uses a dual-branch residual AE to separately extract features from IVI, addressing the challenges of feature extraction discrepancies between the two modalities. Secondly, this paper introduces a dynamic cross-attention fusion network (DCAFNet) to deeply explore the spatial correlations between IVI features, enabling higher-quality fusion. Thirdly, the fused features are used as condition to guide the diffusion models, enhancing the fusion effect while improving cross-scene generalization. Finally, a multi-level dense-connected decoder (MDD) is proposed to generate the fused image. MDD allows deeper global information to complement shallow features better through dense connections, thereby enhancing the feature representation capability of the image. Extensive experiments on several public datasets show that our method achieves superior fusion performance and generates fused images better than existing state-of-the-art methods. Specifically, D2Fusion achieved EN (6.9131), SD (43.0563), and SSIM (0.9921) on the MSRS dataset.

In foggy environments, infrared images suffer from reduced contrast, degraded details, and blurred objects, which impair detection accuracy and real-time performance. To tackle these issues, we propose IDDNet, a lightweight infrared object detection network that integrates multi-scale fusion dehazing. IDDNet includes a multi-scale fusion dehazing (MSFD) module, which uses multi-scale feature fusion to eliminate haze interference while preserving key object details. A dedicated dehazing loss function, DhLoss, further improves the dehazing effect. In addition to MSFD, IDDNet incorporates three main components: (1) bidirectional polarized self-attention, (2) a weighted bidirectional feature pyramid network, and (3) multi-scale object detection layers. This architecture ensures high detection accuracy and computational efficiency. A two-stage training strategy optimizes the model’s performance, enhancing its accuracy and robustness in foggy environments. Extensive experiments on public datasets demonstrate that IDDNet achieves 89.4% precision and 83.9% AP, showing its superior accuracy, processing speed, generalization, and robust detection performance.

Image dehazing helps visual systems adapt to changing weather conditions. Hazy environments significantly reduce image clarity and contrast, which can impair high-level vision tasks such as object detection. To address this issue, this paper proposes a lightweight model for visible image dehazing called the Cross-Fusion Dehazing Model (CFDM). The model restores image clarity through cross-fusion and residual connections. A dynamic dehazing module (DDM) is integrated into CFDM to remove haze while preserving image features. DDM consists of lightweight depthwise separable convolution (DSConv), a dynamic Tanh layer (DyT), and the GeLU activation function. This design improves efficiency while enhancing adaptability to various hazy scenes. In addition, this paper adopts an adaptive fusion strategy to integrate features from different layers. This ensures more effective feature fusion during transmission and facilitates subsequent modules in processing multi-level information. Experimental results show that the proposed model performs well in improving image clarity, preserving details, and removing haze. In addition, the model has a simple structure, low parameter count, and fast inference speed, making it practical and efficient.

In today’s research, unimodal target detection can no longer meet the needs of target detection in complex backgrounds as well as harsh environments. To solve the problems of the existing cross-modal image fusion and cross-modal image target detection algorithms with network-heavy parameters and redundant network design, a selectable cross-modal image fusion and target detection algorithm framework (CMIFDF) is proposed. The framework consists of a lightweight dual-branch cross-modal image fusion network (LDFnet) and a cross-modal object detection algorithm with shareable weights (CM-YOLO) to rationally utilize the cross-modal image information and improve the performance of target detection under complex backgrounds. LDFnet is a two-branch fusion module based on depth-separable convolutional and attentional mechanisms. It can quickly and fully extract feature information from visible and infrared images. In CM-YOLO, fused images or raw images (visible and infrared) are fed into a target detection network with shareable weights for training and detection. A simplified asymptotic feature pyramid network (SAFPN) is proposed, and a lightweight multilayer perceptual attention module (LMA) is designed to enhance the fusion efficiency of the fusion network, so that efficient fusion of features can be achieved with fewer model parameters and low dissipation power to improve the network detection performance. Experiments on publicly available datasets show that the algorithmic framework can make full use of the feature information of cross-modal images as inputs and can effectively improve detection performance in complex environments.

Increasingly massive image data is restricted by conditions such as information transmission and reconstruction, and it is increasingly difficult to meet the requirements of speed and integrity in the information age. To solve the urgent problems faced by massive image data in information transmission, this paper proposes a high-quality image compression algorithm based on unsupervised learning. Among them, a content-weighted autoencoder network is proposed to achieve image compression coding on the basis of a smaller bit rate to solve the entropy rate optimization problem. Binary quantizers are used for coding quantization, and importance maps are used to achieve better bit allocation. The compression rate is further controlled and optimized. A multi-scale discriminator suitable for the generative adversarial network image compression framework is designed to solve the problem that the generated compressed image is prone to blurring and distortion. Finally, through training with different weights, the distortion of each scale is minimized, so that the image compression can achieve a higher quality compression and reconstruction effect. The experimental results show that the algorithm model can save the details of the image and greatly compress the memory of the image. Its advantage is that it can expand and compress a large number of images quickly and efficiently and realize the efficient processing of image compression.

This paper focuses on the handover position and terminal guidance design of a certain type of short-range axisymmetric air-to-ground missile. To guide the missile to the vicinity of the target area for target search, a performance-based evaluation function is established to comprehensively optimize the handover position. To meet the constraints of terminal impact angle, a comparative analysis is conducted between the sliding mode guidance law and the over-gravity compensation proportional guidance law, as well as the ballistic shaping guidance law. The results show that the sliding mode guidance law is more efficient in the context of this paper. Finally, a two-loop autopilot is established to obtain a complete guidance control system for simulation, and the effectiveness of the scheme is verified using Monte Carlo method simulation.

The detection of infrared moving small objects faces significant challenges in the field of object detection for air vehicles. These types of objects usually occupy a small number of pixels in an infrared image, resulting in limited feature information, considerable feature loss, low recognition accuracy, and various challenges in single-frame detection. To address these challenges, this paper proposes an efficient multi-input method named Multi-YOLOv8, which is based on the YOLOv8s model. The proposed method uses current frames as a primary input and incorporates optical flow processing images and background suppression images as auxiliary inputs to improve detection performance. In addition, an improved method is developed for optical flow computations, named the pyramidal weight-momentum Horn–Schunck (PWMHS) method, which can process optical flows efficiently and precisely. An improved version of the Wise-IoU (WIoU) v3, referred to as α⁎-WIoU v3, is proposed as a bounding box regression (BBR) loss function to optimize the YOLOv8 network. Further, the BiFormer module and lightweight convolution GSConv are introduced to improve the attention to key information for the objects and balance the computational cost and detection performance, respectively. Moreover, a small object detection layer is added the YOLOv8 network to improve the capability for small object detection. Finally, a warming-up training method that can reduce the dependency on auxiliary inputs and ensure model stability in case of auxiliary input failures is developed. The results of the comprehensive experiments on an open-access dataset reveal that the proposed model outperforms the mainstream models in overall performance. The proposed method can significantly enhance the detection ability of infrared moving small objects.

AbstractTransformer possesses a broader perceptual scope, while the Convolutional Neural Network (CNN) excels at capturing local information. In this paper, the authors propose the Multi‐Sclale Feature Aggregation Network (MSFA‐Net) for single‐image dehazing which is fused with the advantages of Transformer and CNN. Our MSFA‐Net is based on the encoder–decoder structure, and there are four main innovations. Firstly, the authors make some improvements to the original Swin Transformer to make it more effective for dehazing tasks, and the authors name it Spatial Information Aggregation Transformer (SIAT). The authors place the SIAT in both encoder and decoder of MSFA‐Net for feature extraction. The authors propose an upsampling module called Efficient Spatial Resolution Recovery (ESRR) which is placed in the decoder part. Compared to commonly used transposed convolutions, the authors’ ESRR module has fewer computational cost. Considering that the haze distribution is always uneven and the information from each channel is different, the authors introduce the Dynamic Multi‐Attention (DMA) module to provide pixel‐wise weights and channel‐wise weights for input features. The authors place the DMA module between the encoder and decoder parts. As the network depth increases, the spatial structural information from the high‐resolution layer tends to degrade. To deal with the problem, the authors propose the Multi‐Scale Feature Fusion (MSFF) module to recover missing spatial structural information. The authors place the MSFF module in both the encoder and decoder parts. Extensive experimental results show that the authors’ proposed dehazing network achieves state‐of‐the‐art dehazing performance with relatively low computational cost.

Infrared images of sensitive targets are difficult to obtain and cannot meet the design and training needs of target detection and tracking algorithms for mobile platforms such as aircraft. This paper proposes an image translation algorithm TransImg, which can achieve visible light image translation to the infrared domain to enrich the dataset. First, the algorithm designed a generator structure consisting of a deep residual connected encoder and a region perception feature fusion module to enhance feature learning, thereby avoiding issues such as generating infrared images with insufficient details in the transfer task. Afterward, a multi-scale discriminator and a composite loss function were designed to further improve the transfer effect. Finally, an automatic mixed-precision training strategy was designed for the overall migration algorithm architecture to accelerate the training and generation of infrared images. Experiments have shown that the image translation algorithm TransImg has good algorithm accuracy, and the infrared image generated by visible light image translation has richer texture details, faster generation speed, and lower video memory consumption, and the performance exceeds the mainstream traditional algorithm, and the generated images can meet the requirements of target detection and tracking algorithms design and training for mobile platforms such as aircraft.

In recent years, cross-modal object detection has attracted much attention from researchers across various domains. Compared to single-modal detection, cross-modal object detection combines diverse features from distinct modalities, bolstering the reliability and robustness of object detection applications. However, presently proposed cross-modal models exhibit deficiencies in terms of fusion methodologies, resulting them less suitable for the specific task of aerial object detection.To address these challenges, we present the Cross-Modal Adaptive Fusion Network based on Illumination Awareness (CAFN-IA). The Illumination Awareness module is designed to dynamically adjust the position and orientation of GroundTruth-Box (GT-Box) by quantifying light intensity and computing Intersection over Union (IoU) metrics derived from RGB and Infrared images. Additionally, a dual-stream network architecture is developed to extract RGB and Infrared features separately. Moreover, the introduction of an Interest Region Extraction module enhances the extraction of partial regions. Furthermore, we introduce a Cross-Scale Adaptive Fusion module, enhancing the complementarity of distinct features generating from RGB and Infrared images. Notably, our approach involves the modification of the loss function to elevate the accuracy of small object detection.Extensive experimentation and thorough ablation studies demonstrate the efficacy of our method, yielding an accuracy rate surpassing 69% on the DroneVehicle Datasets.

Aiming at improving the control precision and stability of attack guidance control law against fixed targets and low-speed mobile targets. Combined with the sliding mode control theory, this paper first transforms the problem with angle of impact constraint into a sliding mode surface design problem with LOS angle and LOS rate as state space variables, and then carries out a joint design with the reaching law, and finally puts forward the terminal sliding mode attack guidance law. The stability and convergence of the guidance law are proved by the Lyapunov stability mathematical principle. The theoretical design and experimental data show that compared with the traditional guidance control law, the attack guidance law proposed in this paper can achieve good fall angle constraint and extremely high hit rate design, and have the advantages such as fast convergence and high robustness.

To improve the accuracy and speed of missile detection and tracking for infrared moving small objects (IMSO), a continuous detection and tracking algorithm called CDT-YOLO is proposed. The algorithm mainly consists of an object detection module based on an improved YOLOv8, a tracking confidence evaluation module, and a multi-scale tracking module based on template matching. The object detection module has demonstrated excellent detection performance for IMSO in complex backgrounds. The YOLOv8 model introduces a lightweight convolution known as GSConv to balance detection accuracy and speed. Additionally, the model uses Wise-IoU (WIoU) v3 as a bounding box regression loss to improve localization ability. The object tracking module shows high robustness to size transformation and rapid movement of small objects. The tracking confidence evaluation module is able to re-invoke the detection algorithm to resume tracking the object when the tracking status is found to be abnormal. After experimental validation, the inference speed of the object detection algorithm exceeds 134 FPS with precision up to 0.842 and AP50 up to 0.692. The CDT-YOLO algorithm outperforms classic tracking algorithms in terms of tracking success rate (Sr 60 ) up to 0.774 and speed more than 290 FPS. The method also has the capability of long-term tracking and can be operated in real-time on the missile platform.