A5100422747- Industrial Vision Systems and Defect Detection
- Image Processing Techniques and Applications
- Optical measurement and interference techniques
- Manufacturing Process and Optimization
- Advanced Measurement and Detection Methods
- Textile materials and evaluations
- Advanced Sensor and Energy Harvesting Materials
- Image and Object Detection Techniques
- Solar and Space Plasma Dynamics
- ECG Monitoring and Analysis
- EEG and Brain-Computer Interfaces
- Color Science and Applications
- Advanced SAR Imaging Techniques
- Conducting polymers and applications
- Tactile and Sensory Interactions
- Vehicle License Plate Recognition
- Ionosphere and magnetosphere dynamics
- Cardiac electrophysiology and arrhythmias
- 2D Materials and Applications
- Non-Invasive Vital Sign Monitoring
- Optical Systems and Laser Technology
- Image and Video Stabilization
- Design Education and Practice
- Nanowire Synthesis and Applications
- Surface Roughness and Optical Measurements
Hebei University
2023-2025
Northeastern University
2012-2025
Beihang University
2011-2024
South China University of Technology
2012-2024
Panzhihua University
2022-2023
Xi’an University of Posts and Telecommunications
2012-2023
Foton Motors (China)
2023
Beijing Luhe Hospital Affiliated to Capital Medical University
2022
Dalian University of Technology
2021
Universidad del Noreste
2019-2020
Abstract Oxide materials with a non-centrosymmetric structure exhibit bulk photovoltaic effect (BPVE) but low cell efficiency. Over the past few years, relatively larger BPVE coefficients have been reported for two-dimensional (2D) layers and stacks asymmety-induced spontaneous polarization. Here, we report crucial breakthrough in boosting 3R-MoS 2 by adopting edge contact (EC) geometry using bismuth semimetal electrode. In clear contrast to typically used top (TC) geometry, EC metal which...
The application of license plate recognition technology is becoming more and extensive. In view the current practical requirements for accuracy real-time performance system in complex scenes, existing target detection methods are studied, a car method based on improved YOLOv5m LPRNet model proposed. On basis studying algorithm image features plate, from three aspects: K-means++ used to improve matching degree between anchor frame target, DIOU loss function NMS method, feature map with 20×20...
Three different Si/CdS heterojunction position-sensitive detectors with planar, pyramid, and nanowire structures are successfully prepared systematically investigated. Among these configurations, the core-shell structure demonstrates superior self-powered capability exceptional lateral photovoltaic effect (LPE) performance, achieving a broad optical response from 405 to 1064 nm, peak positional sensitivity of 779.1 mV/mm, minimal nonlinearity just 6.3%. Notably, introducing pyroelectric...
As there are error judgments of float type in the traditional method based on image processing, it is hard to determine woven fabric pattern from recognition results. To solve this problem, fuzzy C‐means (FCM) algorithm was selected classify floats into two groups experiment, and BP neural network chosen recognize pattern. White–black co‐occurrence matrix used extract its texture features. The structure features normal fabrics extracted classification input complete learning process. During...
In this paper, the development of FPGA-based ANNs is presented. Field Programmable Gate Arrays (FPGA) based Artificial Neural Network (ANN) now becoming a focus ANN research. According to parallelism and nature features (NN), hardware implementation superior comparing with software approach because it can take advantage these characteristics. Furthermore, since FPGA digital device that owns reprogrammable properties robust flexibility, many researchers have made great efforts on realization...
Electrocardiogram (ECG) is a physiological signal widely used in monitoring heart health, which of great significance to the detection and diagnosis diseases. Because abnormal rhythms are very rare, most ECG datasets have data imbalance problems. At present, many algorithms for anomaly automatic recognition affected by imbalance. Conventional augmentation methods not suitable signal, because one-dimensional their morphology has significances. In this paper, we propose ProGAN based sample...
In the process of analyzing yarn-dyed fabric, two kinds color information about yarns should be detected: (1) number yarn colors; (2) layout yarns. The traditional detection methods are time-consuming and labor-intensive. An automatic method based on image analysis is proposed in this study. fabric captured with a flat scanner analyzed by fuzzy C-means clustering (FCM) algorithm. By FCM algorithm, we can conclude that colors obtained cluster validity analysis, inspected automatically help...
Fast sampling of wideband RF signal and real time processing vast data are great difficulties in the engineering implementation deception jamming to ISAR. This paper proposes a sort Sub-Nyquist Sampling-Based Method for Jamming ISAR Systems. First introduces technique based on false target Then investigates principle sub-Nyquist modulation. Through modulating information radar signals intercepted by sampling, resulting can produce train vivid targets. The proposed method greatly reduce...
A large amount of seam detection for inhomogeneously textured fabrics makes those performing it fatigued, which leads to misjudgments by human vision, especially the patterned fabrics. The traditional wavelet texture analysis is no longer applicable with inhomogeneous textures and irregular patterns. In this paper, a novel mean weighting factor proposed obtain an optimized discriminant measure detect fabric seams. Firstly, coefficients are extracted in individual decomposition levels. Then...
The texture of yarn-dyed fabric image has been analyzed in HSL color space. With Fourier Transform, periods the pixels a line are obtained. By comparing different lines, conclusion that if is parallel to yarns, sum for three components (H,S,L) comes minimum, proposed this paper. conclusion, center chosen as skew rectification point. All lines within ± 20 degree traversed with step 0.2 degree, and H,S L these all calculated FFT. each line, 101 values Then position yarns can be located by...
Deep neural networks (DNNs) are currently the best-performing method for many classification problems. For training DNNs, learning rate is most important hyper-parameter, choice of which affects performance model greatly. In recent years, some schedulers, such as HTD, CLR, and SGDR, have been proposed. These methods, make use cycling mechanism to improve convergence speed accuracy DNN, but degradation occurs in process. Others good accuracy, their too slow. This paper proposed a new schedule...
Multi-Source cross-lingual transfer learning deals with the of task knowledge from multiple labelled source languages to an unlabeled target language under shift. Existing methods typically focus on weighting predictions produced by language-specific classifiers different sources that follow a shared encoder. However, all share same encoder, which is updated these languages. The extracted representations inevitably contain languages' information, may disturb classifiers. Additionally, due...
Fluorescence microscopic images of cells contain a large number morphological features that are used as an unbiased source quantitative information about cell status, through which researchers can extract and study the biological phenomena statistical analytical analysis. As important research object phenotypic analysis, have great influence on results. Saturation artifacts present in image result loss grayscale does not reveal true value fluorescence intensity. From perspective data...