This study investigates a polarization-insensitive surface plasmon resonance (SPR) sensor based on a D-shaped photonic crystal fiber (PCF) with a sinusoidally modulated surface morphology. By optimizing structural parameters, including the period, amplitude, and core proximity of the sinusoidal surface profile, we systematically explore their influence on SPR characteristics and polarization-independent performance. The results demonstrate that reduced periods or increased amplitudes effectively achieve polarization independence, manifested by equal X- and Y-polarized loss peaks at identical wavelengths. Numerical simulations reveal polarization-independent operation at periods of 5 µm, 7.5 µm, 10 µm, and 12.5 µm through appropriate amplitude adjustments. Furthermore, the impacts of critical fiber parameters (air hole diameter, pitch, and metal film thickness) on SPR excitation are comprehensively analyzed, accompanied by proposed optimization guidelines. This work provides novel insights into the design of D-shaped PCF-SPR sensors with polarization insensitivity, demonstrating promising potential for diverse sensing applications.

In this work, the intelligent reconfigurable surface (IRS) has been proposed to relax the requirement of line-of-sight (LOS) channel gain of the cell-edge user in the non-orthogonal multiple access visible light communication (NOMA-VLC) system, in which the physical layer security (PLS) of the cell-edge user has been enhanced by optimizing the power allocation method at the transmitter. Besides, different mobility models have been adopted for the paired NOMA users to cope with the real indoor environment. The optimal power allocation method and non-silent probability (NSP) of the transmitter are mathematically derived by considering different cases of users’ channel gains. Through simulations, performances of the secrecy rate of the cell-edge legitimate user and NSP of the transmitter are analyzed in detail, where the impacts of the main system parameters are also studied. This work can serve as a good reference for further study of the secure VLC system.

We propose a star-shaped photonic crystal fiber (PCF) structure with a carbon disulfide (CS₂) core, characterized by flat dispersion and elevated nonlinearity, aimed at facilitating high-coherence octave-spanning supercontinuum generation (SCG). The optical properties of the CS₂-core PCF were investigated through numerical simulations by varying structural parameters. Two optimized CS₂-core PCF structures were identified for achieving optimal bandwidth and high-coherence SCG. The first fiber, designated as #F₁, has a lattice spacing of Λ=1.8 µm pumping with a 100 fs input pulse at 1 554 nm and a power of 1 000 W, a flat octave-spanning SC extending from 1 113 nm to 2 357 nm can be generated. The second fiber, referred to as #F₂, has a lattice spacing of Λ=1.3 µm. By pumping with a 50 fs input pulse at 1 750 nm and 1 500 W, a highly coherent SC ranging from 1 225 nm to 2 350 nm can be generated. This research presents a new PCF design for generating high-coherence octave-spanning SC that can be applied in biosensing, spectroscopy, and other fields.

The presence of suspended solids can significantly impact the accuracy of measuring the chemical oxygen demand (COD) concentration in water quality using the spectral absorption method. To address this issue, this research utilized a chemometric method in conjunction with a high-sensitivity ultraviolet-visible-infrared (UV-VIS-IR) spectrophotometer to conduct spectral analysis in a dark room environment at room temperature. The spectral absorption characteristics of COD and kaolin solutions with varying concentration ratios were systematically explored through cross experiments. The turbidity compensation algorithm was studied in detail by constructing a function of independent and dependent variables to develop a comprehensive model representation. A regression model was constructed using logarithmic function fitting and partial least squares regression (PLSR) algorithms. The research results demonstrate that the methodology yields precise final output results, even in solution with high concentration turbidity. This research presents a departure from the conventional turbidity compensation methodology, which provides a real-time correction and a flexible modeling method for the analysis of adjustment data for COD measurement based on spectral absorbance.

This paper proposes YOLOv5-CJ for steel surface defect detection. C3_MSBlock enhances multi-scale feature extraction and enlarges the receptive field, while DyHead introduces scale-, spatial-, and task-aware attention to improve robustness in complex scenes. Soft non-maximum suppression (NMS) further improves recognition in overlapping regions. Compared with YOLOv5s, YOLOv5-CJ improves the mean average precision at intersection over union (IoU) of 0.5 (mAP0.5) and the mean average precision averaged over IoU threshold from 0.5 to 0.95 (mAP0.5: 0.95) by 1.9% and 7.2% on the Northeastern University steel surface defect (NEU-DET) dataset, and by 5.3% and 4.3% on the GC10 steel surface defect (GC10-DET) dataset, respectively, demonstrating its effectiveness for industrial defect detection.

To suppress the high-level noise of raw images from the low-light image sensor, this paper proposes a collaborative filtering algorithm based on exact noise variance of transform domain. Firstly, the noise of low-light-level images is modeled as Poisson–Gaussian mixed noise and performed by variance stabilizing transformation (VST). Secondly, a calculation method of exact noise variance is proposed based on L1 total generalized variation (L1-TGV) regularization. Finally, the denoised images are obtained by embedding the exact noise variance into block matching and three-dimensional filtering (BM3D) algorithm to improve patch matching and shrinkage accuracy. Numerical experiments on unnaturally degraded images express that the proposed method can effectively remove high-level noise and maintain image textures. Compared with BM3D algorithm, the proposed method can improve the peak signal-to-noise ratio (PSNR) by up to 2.15 dB and the structural similarity (SSIM) by up to 0.106, respectively. Moreover, the testing of the raw low-light images confirms the best performance of vision in contrast with the other four methods.

Existing three-dimensional (3D) face reconstruction methods struggle to capture high-frequency facial details, such as subtle expressions and fine skin textures, essential for accurate reconstruction and realistic user interaction. To address this limitation, we propose the implicit representation fusion network (IRFNet), a novel framework for precise facial geometry reconstruction. IRFNet integrates deformation-aware feature extraction and semantic facial segmentation, effectively combining local and global structural cues to optimize facial geometry accuracy. Additionally, a hybrid feature rendering mechanism enhances reconstruction consistency, particularly in complex environments. Compared to current approaches, IRFNet mitigates the geometric distortions inherent in explicit representations and better adapts to diverse facial morphologies and expression variations. Extensive experiments on real-world facial benchmarks demonstrate that IRFNet achieves state-of-the-art performance in 3D face reconstruction.

The existing multi-label hash retrieval methods fail to adequately capture the fine-grained distinctions between image pairs during similarity evaluations. To address this limitation, we propose a novel fine-grained similarity evaluation method for multi-label image pairs, upon which we develop a fine-grained similarity-based multi-label remote sensing image hash retrieval (FMHR) framework. Specifically, the developed evaluation method establishes hierarchical criteria that systematically account for both common and distinct labels between image pairs. FMHR leverages the proposed evaluation framework to extract multi-dimensional discriminative features from remote sensing images. The experimental results on three public multi-label remote sensing datasets demonstrate that the FMHR approach outperforms other methods in terms of retrieval quality and ranking accuracy.

In order to achieve the evaluation of human rehabilitation training movements, a human 3D pose estimation network integrating key-frame enhancement method (KFEM) and CTRAMM module is proposed, and a matching algorithm based on location and type dynamic time warping (LTDTW) is developed to evaluate rehabilitation movements. KFEM determines key-frames and adjusts their weights by calculating the coordinate transformation of human key-points. The CTRAMM module dynamically learns different topological structures, improving the feature representation ability of the model. The LTDTW improves the accuracy of sequence matching through adaptive weight coefficients. The experimental results on different datasets have validated the effectiveness of the proposed method.

Currently, foundation model had attracted a lot attention for monocular depth estimation in endoscopic surgery. However, degradation in clinical scenarios is often complex for endoscopic image, which leads to compromised robustness. Therefore, we propose a self-supervised generative feature driven EndoMDNet that aims to mitigate degradation for robust monocular depth estimation in endoscopic surgery. Specifically, a content recognition mechanism is designed to guide diffusion model to generate detail information that is utilized as the supplementation of degraded depth feature. Moreover, diffusion model tends to generate artifacts that may also be inconsistent with the target distribution. To address this problem, we propose wavelet-based refined adaptive fusion block (WRAF) to filter noise of generative feature and adaptively fuse it with degraded depth feature. Finally, extensive experiment on screen for child anxiety related emotional disorders (SCARED), stereoscopic endoscopic reconstruction validation-CT (SERV-CT) and Hamlyn datasets demonstrate the robustness of our proposed method.