Yueli Hu

Video denoising aims to recover clean video sequences with rich structural details from temporally correlated noisy frames, where the core challenge lies in effectively aggregating long-range temporal information under complex motion conditions. Existing sliding-window and recurrent-based approaches suffer from limited receptive fields or restricted parallelism, while recent Transformer-based methods, although capable of modeling long-range dependencies, are constrained by fixed-window designs that hinder large-scale motion handling. In this paper, we propose a novel and efficient Shift Alignment Transformer (SAT) for video denoising, which enables implicit feature alignment and aggregation over extended spatiotemporal regions without relying on optical flow estimation. The proposed framework introduces two complementary mechanisms: Time Segment Shift, which establishes inter-frame correlations by dynamically shifting temporal windows, and Local Window Shift, which enhances intra-frame contextual modeling through spatial window displacement. Together, these designs enable flexible receptive field expansion while maintaining computational efficiency. Extensive experiments conducted on both synthetic Gaussian noise removal in the RGB domain and real-world video denoising in the RAW domain demonstrate that SAT consistently outperforms or matches state-of-the-art methods in terms of denoising accuracy, while achieving a favorable balance between performance and computational cost. These results indicate that the proposed SAT provides an effective and practical solution for long-range spatiotemporal modeling in video denoising, offering a robust alternative to explicit motion estimation and fixed-window Transformer architectures.

Currently, infrared object detection is utilized in a broad spectrum of fields, including military applications, security, and aerospace. Nonetheless, the limited computational power of edge devices presents a considerable challenge in achieving an optimal balance between accuracy and computational efficiency in infrared object detection. In order to enhance the accuracy of infrared target detection and strengthen the implementation of robust models on edge platforms for rapid real-time inference, this paper presents LKD-YOLOv8, an innovative infrared object detection method that integrates YOLOv8 architecture with masked generative distillation (MGD), further augmented by the lightweight convolution design and attention mechanism for improved feature adaptability. Linear deformable convolution (LDConv) strengthens spatial feature extraction by dynamically adjusting kernel offsets, while coordinate attention (CA) refines feature alignment through channel-wise interaction. We employ a large-scale model (YOLOv8s) as the teacher to imparts knowledge and supervise the training of a compact student model (YOLOv8n). Experiments show that LKD-YOLOv8 achieves a 1.18% mAP@0.5:0.95 improvement over baseline methods while reducing the parameter size by 7.9%. Our approach effectively balances accuracy and efficiency, rendering it applicable for resource-constrained edge devices in infrared scenarios.

A time-varying process with nonlinearity and time lag is the temperature control of pulsing vacuum steam sterilization. In order to achieve efficient and accurate control requirements, conventional PID temperature control algorithms sometimes display slow response speed, severe overshooting, unstable performance, and other challenges that ultimately affect the sterilizing effect. In order to find the ideal steam sterilization temperature control settings iteratively, this research used the PSO algorithm. Simulating and analyzing the system model is done simultaneously using fuzzy control of the PID parameter adaptive modification. According to the results, there is no overshooting and the response speed approach is faster. This paper presents an approach to fuzzy PID control based on the PSO optimization algorithm. As a result of fuzzy adaptive PID's high control accuracy and quick response time, the PID parameters are also continuously optimized utilizing the PSO approach for steam sterilization temperature control. For the purpose of doing simulation analysis, create and modify a system model. As evidenced by the results, this strategy has a reduced overshoot, a faster response time, and better stability. It may also successfully boost the control effect. Eventually, this method was applied to a self-tuning PID control experiment for sterilizer temperature control, and a relatively optimal control effect was obtained.

The paper provided a brief analysis of video denoising characteristics, discussed and analyzed various existing video denoising methods, and proposed a new video denoising algorithm based on bidirectional time fusion and the W-Net architecture, designed to meet the requirements of real-time video denoising. This algorithm effectively combines past and future information, increases the temporal receptive field, and reduces memory usage. Additionally, by selecting a deeper W-Net backbone network, the algorithm achieves high-fidelity real-time video denoising. Comparative analysis with other video denoising models demonstrated that this approach outperforms others in terms of fidelity.

Video frame interpolation involves synthesizing intermediate frames between two consecutive frames to enhance the smoothness of a video. Nevertheless, generating high-quality interpolated frames in videos featuring substantial motion and complex scenes remains a formidable challenge. To produce superior quality frames, this paper introduces an interpolation method based on multi-scale features and correlation volumes. The multi-scale feature connects the deep features of high-resolution frames with the shallow features of lower-resolution frames, thereby increasing the number of available pixels and feature details for motion analysis. Correlation volumes are employed to construct correlation features for all pairs of pixels, which are utilized to refine the underlying optical flow field. We propose a unified network approach that eliminates the need for additional complex optical flow network integration, simplifying the training process. The experimental results demonstrate that this method outperforms the baseline approach in both objective and subjective evaluations across various datasets. Particularly, this method exhibits advantages on datasets characterized by complex backgrounds and large motions.

Infrared sensors capture infrared rays radiated by objects to form thermal images. They have a steady ability to penetrate smoke and fog, and are widely used in security monitoring, military, etc. However, civilian infrared detectors with lower resolution cannot compare with megapixel RGB camera sensors. In this paper, we propose a dynamic attention mechanism-based thermal image super-resolution network for infrared sensors. Specifically, the dynamic attention modules adaptively reweight the outputs of the attention and non-attention branches according to features at different depths of the network. The attention branch, which consists of channel- and pixel-wise attention blocks, is responsible for extracting the most informative features, while the non-attention branch is adopted as a supplement to extract the remaining ignored features. The dynamic weights block operates with 1D convolution instead of the full multi-layer perceptron on the global average pooled features, reducing parameters and enhancing information interaction between channels, and the same structure is adopted in the channel attention block. Qualitative and quantitative results on three testing datasets demonstrate that the proposed network can superior restore high-frequency details while improving the resolution of thermal images. And the lightweight structure of the proposed network with lower computing cost can be practically deployed on edge devices, effectively improving the imaging perception quality of infrared sensors.

Color constancy is used to determine the actual surface color of the scene affected by illumination so that the captured image is more in line with the characteristics of human perception. The well-known Gray-Edge hypothesis states that the average edge difference in a scene is achromatic. Inspired by the Gray-Edge hypothesis, we propose a new illumination estimation method. Specifically, after analyzing three public datasets containing rich illumination conditions and scenes, we found that the ratio of the global sum of reflectance differences to the global sum of locally normalized reflectance differences is achromatic. Based on this hypothesis, we also propose an accurate color constancy method. The method was tested on four test datasets containing various illumination conditions (three datasets in a single-light environment and one dataset in a multi-light environment). The results show that the proposed method outperforms the state-of-the-art color constancy methods. Furthermore, we propose a new framework that can incorporate current mainstream statistics-based color constancy methods (Gray-World, Max-RGB, Gray-Edge, etc.) into the proposed framework.

Finger veins are widely used in various fields due to their high safety. Existing finger vein recognition methods have some shortcomings, such as low recognition accuracy and large model size. To address these shortcomings, a multi-scale feature bilinear fusion network (MSFBF-Net) was designed. First, the network model extracts the global features and local detail features of the finger veins and performs linear fusion to obtain second-order features with richer information. Then, the mixed depthwise separable convolution replaces the ordinary convolution, which greatly reduces the computational complexity of the network model. Finally, a multiple attention mechanism (MAM) suitable for finger veins was designed, which can simultaneously extract the channel, spatial, directional, and positional information. The experimental results show that the method is very effective, and the accuracy of the two public finger vein databases is 99.90% and 99.82%, respectively.