Jian Qiu

Accurate measurement of the Zeta potential is crucial for evaluating the dispersion and aggregation behavior of colloidal systems. Currently, the prevailing Electrophoretic Light Scattering (ELS) method is an indirect average measurement technique that lacks visual intuitiveness. Meanwhile, traditional micro-electrophoresis is constrained by the optical diffraction limit and fails to account for measurement deviations induced by the axial motion of particles. To enhance the measurement accuracy and visualization characterization capability, this paper proposes a method for visualizing the Zeta potential of nanoparticles. By integrating the correlation operation of dark-field microscopic images and the axial motion speed compensation strategy, the three-dimensional velocity of the particles is measured, and the electrophoretic mobility is calculated using the Henry's equation. Comparative experiments utilizing particles of various sizes demonstrate that the measurement results of the proposed system are highly consistent with those of commercial ELS instruments (relative error < 7%). Furthermore, this method exhibits better applicability in the measurement of extremely low-concentration samples and accurately reflects the variation trends of the Zeta potential in response to pH changes. This strategy combines visualization, high precision and stability, providing a universal dark-field microscopic imaging and measurement platform for the characterization of nanoparticle Zeta potential.

Traditional Visual-Inertial Odometry (VIO) systems degrade in weakly textured and dynamically changing indoor environments, limiting reliable robot localization. To address this issue, we propose Dynamic Structural Visual-Inertial Odometry (Dynamic-SVIO), an indoor Red-Green-Blue and Depth (RGB-D) VIO system that integrates point-line-plane features with the Atlanta World (AW) constraint to enhance geometric observability. Dynamic-SVIO further introduces a dynamic feature identification and removal module that combines object detection and epipolar constraints to suppress misleading observations caused by human motion. A tightly coupled sliding-window backend fuses visual, depth, and inertial measurements for accurate state estimation. On the OpenLORIS dataset, Dynamic-SVIO achieves an average Relative Pose Error (RPE) of 0.030 and an average Absolute Trajectory Error (ATE) of 0.187, corresponding to approximately 25 percent and 19 percent improvements over representative baselines, respectively, and more than 60 percent ATE reduction compared with VINS-Mono. These results demonstrate that jointly leveraging structural priors and dynamic feature suppression significantly improves indoor ego-motion estimation, supporting robust embodied perception in human-robot coexisting environments.

Existing 3D Gaussian Splatting (3DGS)-based Simultaneous Localization and Mapping (SLAM) methods improve tracking accuracy and rendering quality by incorporating additional components such as loop closure. However, this improvement comes at the cost of reduced speed, and these methods have limited the use of image information. In this paper, we propose EAGS-SLAM, an edge-assisted RGB-D Gaussian Splatting SLAM that leverages edge information to improve system performance. EAGS-SLAM combines edge-based visual odometry and gaussian tracking, which reduces camera tracking time by up to 50%. We propose a novel gaussian seeding method that directly utilizes edge information to capture geometric details in images. A decoupled submap design that can run loop closure in parallel, ensuring globally consistent mapping while improving system speed. The proposed system is evaluated on real-world datasets Scan-Net, TUM RGB-D, and synthetic dataset Replica. Compared with existing state-of-the-art systems, the proposed system demonstrates competitive or superior performance in tracking, rendering, and runtime. The code and visualization results are available at https://endermands.github.io/EAGSSLAM.

In electromagnetic compatibility analysis and non-destructive fault diagnosis of circuits, accurately measuring the radio-frequency current distribution in printed circuit board (PCB) is of critical importance. However, for circuit modules without dedicated test points, non-destructive and direct measurement of in-situ current remains a formidable challenge. Currently, a common alternative approach involves measuring the magnetic field directly above the PCB with a magnetic field probe, and then reconstructing the current distribution on the PCB surface using the Biot-Savart law. Nevertheless, this inversion algorithm typically assumes quasi-static current conditions, and its inversion parameters are often empirically selected. These limitations significantly constrain its applicability in complex radio-frequency (RF) environments. To address this challenge, this work proposes an enhanced Magnetic Current Imaging method specifically designed for reconstructing the RF current distribution on PCBs surfaces. This method leverages the current-magnetic field relationship and employs a regularization technique to constrain current distribution. Additionally, it utilizes a mask-based approach and an extended measurement window to mitigate current artifacts. Finally, its feasibility and accuracy are comprehensively validated through simulations and near-field scanning experiments, demonstrating its robustness and effectiveness even under high-noise conditions.

Objective. This study aims to enhance the accuracy and reliability of imaging photoplethysmography (IPPG) for heart rate (HR) measurements during nighttime by introducing an innovative approach that combines fast independent component analysis (FastICA) with a Time-Delayed Multi-Dimensional Extended Regions of Interest Extraction (TDMDE-ROI-Ex) technique, specifically tailored to overcome the challenges posed by motion artefacts and the difficulty in identifying regions of interest (ROIs). Approach. This research employs a dual-method strategy for the precise extraction of ROIs and robust processing of HR signals in nighttime IPPG scenarios. Initially, a face detection algorithm is integrated with a grayscale clustering technique to pinpoint optimal ROIs. This is followed by the application of the mutual information delay method to synthesize multi-channel IPPG signals. Concurrently, the HR’s Fundamental Frequency is leveraged as a prior Constraint within the iterative process of FastICA (HRFFC-FastICA), mitigating the susceptibility to initial value fluctuations inherent in FastICA. The synergistic application of these methodologies substantially bolsters the stability and robustness of nighttime HR measurements, particularly in conditions characterized by significant motion. Main results. The efficacy of the proposed method, which incorporates HRFFC-FastICA, is initially validated through performance testing using the MR-NIRP dataset. This step serves to assess the practicality of the approach for nighttime IPPG HR measurements. Following this, a series of modular ablation studies and comparative evaluations against current nighttime IPPG algorithms are executed. The statistical outcomes demonstrate that our method achieves a mean absolute error (MAE) of 4.57 beats per minute (bpm) and a root mean squared error (RMSE) of 5.95 bpm. In direct comparison with prominent algorithms such as SparsePPG and PhysNet, the method exhibits a notable enhancement in MAE by up to 8.39 bpm and a significant decrease in RMSE by 17.83 bpm. The 95% confidence interval of the Bland–Altman graph of this method is between 9.5 and −12.8 bpm. Compared to other comparable methods, this interval is significantly narrower, with a width nearly half that of alternative approaches, indicating superior precision and reliability. Significance. The significance of this research is highlighted by the experimental outcomes that demonstrate the considerable advantages of the TDMDE-ROI-Ex method. This technique significantly reduces reliance on facial motion, which is crucial for accurately identifying facial skin colour regions of interest. Moreover, integrating the HRFFC-FastICA method effectively counteracts the effects of motion artefacts and the initial value sensitivity inherent in the FastICA process. The introduction of this methodology into nighttime IPPG monitoring significantly strengthens the system’s robustness and stability, thereby extending the range of IPPG technology applications and improving its overall measurement performance.

The development of artificial intelligence and big data has put forward higher requirements for computing devices, and metasurface-based optical devices have attracted attention due to their optical advantages.With the increase in data capacity and the trend of high integration, optical devices are beginning to develop towards multi-dimensional devices, among which wavelength-multiplexing is a method. Here, we propose a spatial differential metasurface with a honeycomb photonic structure as the basic unit to deal with wavelength-multiplexed optical fields, where differential operation is simultaneously performed on two-dimensional (2D) spatial optical fields with incident wavelengths of 1344nm and 1508nm. To verify the effectiveness of the wavelength-multiplexed spatial differentiator, the optical transfer function of the metasurface in one-dimensional (1D) space and 2D space at different wavelengths are numerically simulated and applied to the 1D Gaussian function and 2D image, respectively. The proposed metasurface holds important reference value for the implementation of wavelength-multiplexed differential operators, which provides new opportunities for multi-dimensional optical computing.

Objective . The imaging photoplethysmography (IPPG), which is applied to heart rate measurement, is susceptible to interference caused by illumination variations during photography. To improve the accuracy, robustness, and real-time performance of IPPG, a new method combined with modified ensemble empirical mode decomposition (EEMD) and fast independent component analysis (FastICA) was proposed. Approach . On the modified EEMD, to reduce the error caused by the residual white noise of the algorithm, it was characterized by recurrently updating zero-mean white noise for ensemble averaging based on the standard deviation of the input signal. On the modified FastICA, its constructed nonlinear function was superseded by Huber’s approximation function to improve the robustness and running speed. Main results . Comparison experiments were conducted between the MAHNOB-HCI database and own collected data. In the hybrid natural light and computer screen light scenario, the mean absolute error (MAE) of heart rate amounted to 0.93 beats per minute and the correlation coefficient r reached 0.85. In the experiments of MAHNOB-HCI database, the MAE amounted to 6.03 beats per minute and the correlation coefficient r reached 0.75. Furthermore, the modified method decreased the running time by approximately ten times compared to the original algorithms. Significance . Various experimental results demonstrated that the proposed method significantly improves the accuracy, timeliness, and interference resistance.

—It is important for schools and online learning platforms to mine student attitudes and opinions from student reviews so teachers know what needs to be improved. Previous studies have only focused on the semantic features of sentences, ignoring syntactic dependency relations in sentences. To address this problem, we build a graph convolutional network (GCN) on the syntactic dependency tree of sentences to exploit syntactic information for the first time in the field of online student reviews. Furthermore, in order to exploit positional relations and alleviate over-smoothing, we add relative aspect distance relations and stochastic edge elimination. On this basis, a novel multi-channel graph convolution model (MCGCN) is proposed, which constructs three GCN channels for syntactic dependency relations, relative aspect distance relations and stochastic edge elimination to extract relational features. Finally we use average pooling to fuse three relational features. Experiments on the CR23k dataset demonstrate that the overall performance of the model outperforms the state-of-the-art methods.