In the field of autofocus for optical systems, although passive focusing methods are widely used due to their cost-effectiveness, fixed focusing windows and evaluation functions in certain scenarios can still lead to focusing failures. Additionally, the lack of datasets limits the extensive research of deep learning methods. In this work, we propose a neural network autofocus method with the capability of dynamically selecting the region of interest (ROI). Our main work is as follows: first, we construct a dataset for automatic focusing of grayscale images; second, we transform the autofocus issue into an ordinal regression problem and propose two focusing strategies: full-stack search and single-frame prediction; and third, we construct a MobileViT network with a linear self-attention mechanism to achieve automatic focusing on dynamic regions of interest. The effectiveness of the proposed focusing method is verified through experiments, and the results show that the focusing MAE of the full-stack search can be as low as 0.094, with a focusing time of 27.8 ms, and the focusing MAE of the single-frame prediction can be as low as 0.142, with a focusing time of 27.5 ms.

A variety of factors cause a reduction in remote sensing image resolution. Unlike super-resolution (SR) reconstruction methods with single degradation assumption, multi-degradation SR methods aim to learn the degradation kernel from low-resolution (LR) images and reconstruct high-resolution (HR) images more suitable for restoring the resolution of remote sensing images. However, existing multi-degradation SR methods only utilize the given LR images to learn the representation of the degradation kernel. The mismatches between the estimated degradation kernel and the real-world degradation kernel lead to a significant deterioration in performance of these methods. To address this issue, we design a reconstruction features-guided kernel correction SR network (RFKCNext) for multi-degradation SR reconstruction of remote sensing images. Specifically, the proposed network not only utilizes LR images to extract degradation kernel information but also employs features from SR images to correct the estimated degradation kernel, thereby enhancing the accuracy. RFKCNext utilizes the ConvNext Block (CNB) for global feature modeling. It employs CNB as fundamental units to construct the SR reconstruction subnetwork module (SRConvNext) and the reconstruction features-guided kernel correction network (RFGKCorrector). The SRConvNext reconstructs SR images based on the estimated degradation kernel. The RFGKCorrector corrects the estimated degradation kernel by reconstruction features from the generated SR images. The two networks iterate alternately, forming an end-to-end trainable network. More importantly, the SRConvNext utilizes the degradation kernel estimated by the RFGKCorrection for reconstruction, allowing the SRConvNext to perform well even if the degradation kernel deviates from the real-world scenario. In experimental terms, three levels of noise and five Gaussian blur kernels are considered on the NWPU-RESISC45 remote sensing image dataset for synthesizing degraded remote sensing images to train and test. Compared to existing super-resolution methods, the experimental results demonstrate that our proposed approach achieves significant reconstruction advantages in both quantitative and qualitative evaluations. Additionally, the UCMERCED remote sensing dataset and the real-world remote sensing image dataset provided by the “Tianzhi Cup” Artificial Intelligence Challenge are utilized for further testing. Extensive experiments show that our method delivers more visually plausible results, demonstrating the potential of real-world application.

The point cloud is disordered and unstructured, and it is difficult to extract detailed features. The detailed part of the target shape is difficult to complete in the point cloud completion task. It proposes a point cloud completion network (BCA-Net) focusing on detail reconstruction, which can reduce noise and refine shapes. Specifically, it utilizes residual deformation architecture to avoid error points. The break and recombine refinement method is used to recover complete point cloud details. In addition, it proposes a bilateral confidence aggregation unit based on recurrent path aggregation to refine the coarse point cloud shape using multiple gating. Our experiments on the ShapeNet and Complete3D datasets demonstrate that our network performs better than other point cloud completion networks.

Spatial infrared spot target (SIST) discrimination based on infrared radiation sequences (IRSs) can be considered a univariate trending time series classification task. However, due to the complexity of actual scenarios and the limited opportunities for acquiring IRSs, resulting in noise interference, extremely small-scale datasets with imbalanced distribution of classes and widely varying sequence lengths range from a few hundred to several thousand time steps. Current research is primarily based on idealized simulation datasets, resulting in a performance gap when applied to actual applications. To address these issues, firstly, we construct a simulation dataset tailored to the challenges of actual scenarios. Secondly, we design a practical data preprocessing method to achieve uniform sequence length, coarse alignment of shapelets and filtering while preserving key points. Thirdly, we propose a residual network Res-LK-SLR for IRS classification based on large kernels (LKs, providing long-term dependence) and shapelet-level representations (SLRs, where the hidden layer features are aligned with the learned high-level representations to obtain the optimal segmentation and generate shapelet-level representations). Additionally, we conduct extensive evaluations and validations on both the simulation dataset and 18 UCR time series classification datasets. The results demonstrate the effectiveness and generalization ability of our proposed Res-LK-SLR.

Infrared small target detection (IRSTD) is crucial in aerospace and maritime rescue applications. Since imaging devices, such as satellites, are usually at a very long distance from the targets, the targets appear dim in the images. Existing IRSTD methods predominantly employ single-branch architectures with complex feature extraction modules, which paradoxically achieve limited performance improvement as small targets inherently lack sufficient features for effective characterization. To address this issue, we propose a Dual-Branch Dense Collaborative Network (DB-DCNet), which integrates two specialized detection branches: one focuses on minimizing missed detections through target-sensitive feature learning; the other aims to reduce false alarms via clutter-aware optimization techniques. Both branches adopt the same densely connected network structure but use different feature extraction modules. One branch leverages small-sized kernels and hybrid attention mechanisms to segment target edges accurately, thus effectively decreasing missed detections. On the contrary, the other branch uses convolutions with multiple receptive fields and residual connections to filter out background clutter, significantly suppressing false alarms. The detection results of the two branches are further processed by a lightweight feature selection network for feature selection and fusion, leading to output results with fewer false alarms and missed detections. Comprehensive experiments on the Earth observation datasets NUDT-SIRST, IRSTD-Air, IRSTD-1k, and WideIRSTD-Full verify the superiority of our proposed method.