A5101957629- Advanced SAR Imaging Techniques
- Radar Systems and Signal Processing
- Visual Attention and Saliency Detection
- Infrared Target Detection Methodologies
- Wireless Signal Modulation Classification
- Advanced Antenna and Metasurface Technologies
- Advanced Image Fusion Techniques
- Electromagnetic wave absorption materials
- Electromagnetic Scattering and Analysis
- Medical Image Segmentation Techniques
- Image Enhancement Techniques
- Surface Roughness and Optical Measurements
- Antenna Design and Analysis
- Spectroscopy and Chemometric Analyses
- Advanced Image and Video Retrieval Techniques
- Image and Object Detection Techniques
- Sparse and Compressive Sensing Techniques
- Non-Invasive Vital Sign Monitoring
- Robotic Path Planning Algorithms
- Spectroscopy Techniques in Biomedical and Chemical Research
- Target Tracking and Data Fusion in Sensor Networks
- Image and Signal Denoising Methods
- Metamaterials and Metasurfaces Applications
- Image Retrieval and Classification Techniques
- Remote-Sensing Image Classification
Harbin Engineering University
2021-2024
Ministry of Industry and Information Technology
2024
Peking University
2015
Naval University of Engineering
2009-2010
Xiaomi (China)
2009
Dense false target jamming (DFTJ) is one of the most common and threatening modes, seriously affecting a radar from detecting target. This article counters DFTJ by proposing fast–slow time domain joint frequency response feature-based recognition method that utilizes differences in characteristics digital radio memory (DRFM) jammer. The core idea this to model influence radar's DRFM jammer's on features echo signal. In addition, developed strategy extends fast-time through pulse compression...
This paper presents a new noninvasive blood glucose monitoring method based on four near infrared spectrums and double artificial neural network analysis. We choose wavelengths, 820 nm, 875 945 1050 as transmission spectrums, capture fingers PPG signals simultaneously. The wavelet transform algorithm is used to remove baseline drift, smooth extract eight eigenvalues of each signal. are the input parameters analysis model. Double regression combines classification recognition with prediction...
目的 伪装目标是目标检测领域一类重要研究对象，由于目标与背景融合度较高、视觉边缘性较差、特征信息不足，常规目标检测算法容易出现漏警、虚警，且检测精度不高。针对伪装目标检测的难点，基于YOLOv5（youonly look once）算法提出了一种基于多检测层与自适应权重的伪装目标检测算法（algorithm for detecting camouflagetargets based on multi-detection layers and adaptive weight，MAH-YOLOv5）。方法 在网络预测头部中增加一个非显著目标检测层，提升网络对于像素占比极低、语义信息不足这类目标的感知能力；在特征提取骨干中融合注意力机制，调节卷积网络对特征信息不足目标的权重配比，使其更关注待检测的伪装目标；在网络训练过程中使用多尺度训练策略，进一步提升模型鲁棒性与泛化能力；定义了用于军事目标检测领域的漏警、虚警指标，并提出伪装目标综合检测指数。结果 实验在课题组采集的伪装数据集上进行训练和验证。结果表明，本文方法在自制数据集上的平均精度均值（mean average...
Full forwarding dense false target jamming signals correlate highly with real echoes, and their training samples are difficult to obtain, constraining radars from effectively identifying echoes. To overcome this challenging problem, article systematically analyzes studies the frequency response characteristics of radar jammer models influence on amplitude–frequency features Then, positive-unlabeled learning (PU learning)-based algorithm is proposed solve signal recognition problem missing...
With the widespread adoption of flexible stealth materials (FSMs) in radar application, understanding their electromagnetic scattering information has become increasingly crucial. However, acquiring properties FSMs solely through measurement can be challenging. Hence, this study proposes an model for based on random undulating units formed by sweeping two parabola functions. Specifically, multi-layer fast multipole method (MLFMM) is applied to calculate model’s monostatic RCS within 8–12...
Abstract To effectively address the challenges posed by intricate and dynamic electromagnetic environments, we propose a wideband tunable radar absorber in this paper. The proposed absorber, composed of graphene capacitor, plasma enclosed within sealed glass cavity, absorbing material (RAM), FR-4 copper plate, allows for performance through manipulation properties capacitor plasma. Based on equivalent circuit model, reflectivity is analyzed using transmission line theory (TLT). Good...
Image segmentation is an important problem on image processing technique. Based active contour model, a novel integrated (IAC) model for SAR imagery has been developed. In this the edge information extracted by detection operators based ratio of average framework, and region statistical distilled maximization likelihood function different regions. And unconditionally stable numerical scheme used means additional operator splitting arithmetic. end, some tests have done using MSTAR, Radarsat-2...
Abstract In order to effectively improve the detection ability of camouflaged targets and fully utilize potential target algorithms, this paper proposes a method based on fusion thermal infrared saliency map visible image. Firstly, target’s band is obtained through instance segmentation network dedicated target, then fused with image pixel-level weighting, thus obtaining same in two bands, aimed at enhancement specific regions target. The by YOLOv8, experimental results show that has...
Radar false target jamming can be used to cover the actions of real targets, posing a great challenge radar detection. Most traditional recognition methods are almost ineffective in recognizing real-false targets case missing samples, and their effectiveness may decrease with pulse compression processing. This letter proposes method based on fast-slow time frequency response features positive-unlabeled (PU) learning algorithm overcome above problems. The novel utilizes coherent accumulation...
Aiming at the problems in current coating surface defects detection that it is difficult to characterize defect features, furthermore accuracy and efficiency are hard meet industrial demand, this paper, a machine vision system for designed; then, classification method based on convolutional neural network which trained tested through cross-validation realize of multi-type defects, proposed. According collected dataset including defect-free four types defects: crack coating, running orange...
A Gamma-distributed texture in coherent imagery is widely applied on some fields such as description of characteristic and verification classification algorithm. In this paper, a method based spherically invariant random process for modeling SAR has been proposed, the feasibility with SIRP testified, representation functions have also educed. By means relationship between envelope complex SIRV, correlation property introduced into texture. And simulating flow correlated using SIRV theorem...
In order to solve the problem that single-background camouflage patterns designed by existing methods cannot meet requirements under multi-terrain background, a multi-background pattern design method based on feature fusion is proposed, and formed complete set of theory. Through similarity analysis background images, color determination, patch extraction other steps, shape features are extracted, then merged generate pattern, which suitable for multi-regional camouflage. The simulation...
There are many types of radar active deception false target jamming that highly correlated with the real target. Recognizing and targets under a low Signal-to-Noise Ratio (SNR) is difficult. To solve above problem, this article proposes real/false recognition method based on features multi-pulse joint frequency response by analyzing differences in scattering characteristics modeling echoes as synthesis multi-scattering center echoes. Firstly, range-doppler domain, truncated along range...
It is an important precondition for military Unmanned Aerial Vehicle (UAV) trajectory planning and mission deduction by modeling the threats faced UAV in battlefield environment. We introduced basic theory of Voronoi diagram environment present based method multirotor drones three-dimensional space. In particularly, we propose a process calculate cost each path on threat level energy consumption with flight altitude information.