- Advanced SAR Imaging Techniques
- Face recognition and analysis
- Radar Systems and Signal Processing
- Face and Expression Recognition
- Direction-of-Arrival Estimation Techniques
- Sparse and Compressive Sensing Techniques
- Domain Adaptation and Few-Shot Learning
- Underwater Acoustics Research
- Machine Learning and ELM
- Facial Nerve Paralysis Treatment and Research
- Image Processing and 3D Reconstruction
- Photoacoustic and Ultrasonic Imaging
- Medical Image Segmentation Techniques
- Image and Signal Denoising Methods
- Advanced Data Compression Techniques
- Ultra-Wideband Communications Technology
- Microwave Imaging and Scattering Analysis
- Advanced Neural Network Applications
- Diverse Topics in Contemporary Research
- Remote-Sensing Image Classification
- Non-Invasive Vital Sign Monitoring
- Advanced Wireless Communication Techniques
- Digital Filter Design and Implementation
- Antenna Design and Optimization
- Wireless Signal Modulation Classification
Wuyi University
2017-2024
Helmut Schmidt University
2024
Wuyi University
2020-2022
Beihang University
2013-2016
Face recognition (FR) with single sample per person (SSPP) is a challenge in computer vision. Since there only one to be trained, it makes facial variation such as pose, illumination, and disguise difficult predicted. To overcome this problem, paper proposes scheme combined traditional deep learning (TDL) method process the task. First, an expanding based on approach. Compared other methods, can used easily conveniently. Besides, generate samples disguise, expression, mixed variation....
Facial beauty prediction (FBP) has become an emerging area in the field of artificial intelligence. However, lacks data and accurate face representation hinder development FBP. Multi-task transfer learning can effectively avoid over-fitting, utilize auxiliary information related tasks to optimize main task. In this paper, we present a network named Multi-input Beauty Network (2M BeautyNet) use predict facial beauty. experiment, is task, gender recognition auxiliary. For multi-task training,...
Since Synthetic Aperture Radar (SAR) targets are full of coherent speckle noise, the traditional deep learning models difficult to effectively extract key features and share high computational complexity. To solve problem, an effective lightweight Convolutional Neural Network (CNN) model incorporating transfer is proposed for better handling SAR recognition tasks. In this work, firstly we propose Atrous-Inception module, which combines both atrous convolution inception module obtain rich...
Abstract Facial beauty prediction (FBP) is an important and challenging problem in the fields of computer vision machine learning. Not only it easily prone to overfitting due lack large-scale effective data, but also difficult quickly build robust facial evaluation models because variability appearance complexity human perception. Transfer Learning can be able reduce dependence on large amounts data as well avoid problems. Broad learning system (BLS) capable completing building training. For...
This study presents new insights into sparse frequency waveform analysis and design methods. For a general radar waveform, the total ambiguity in its auto‐correlation function (ACF) is equal to energy power spectral density (PSD) domain. With this relationship, an ACF found be minimised when corresponding PSD uniformly distributed. property extended by establishing relationship between optimal minimum ACF. Based on analysis, method of designing with sidelobe constraint proposed. The...
Synthetic Aperture Radar (SAR) Automatic Target Recognition (ATR), most algorithms of which have employed and relied on sufficient training samples to receive a strong discriminative classification model, has remained challenging task in recent years, among the challenge SAR data acquisition further insight into intuitive features images are main concerns. In this paper, deep transferred multi-level feature fusion attention network with dual optimized loss, called (MFFA-SARNET), is proposed...
In recent years, impressive performance of deep learning technology has been recognized in synthetic aperture radar (SAR) automatic target recognition (ATR). Since a large amount annotated data are required this technique, it poses trenchant challenge to the issue obtaining high rate through less labeled data. To overcome problem, inspired by contrastive learning, we proposed novel framework named batch instance discrimination and feature clustering (BIDFC). framework, different from that...
Multiple-input multiple-output (MIMO) radar equipped with a frequency diverse array (FDA) can produce range-dependent beampattern and increase the degrees-of-freedom of antenna array. In this paper, new method designing MIMO sparse waveforms is proposed for FDA, which randomly samples multiple distance points such that both spectrum constant modulus constraints are realized by design framework. The main steps as follows. We first obtain covariance matrix transmitted signal given ideal...
Multiple-input multiple-output (MIMO) radar takes the advantages of high degrees freedom for beam pattern design and waveform optimization, because each antenna in centralized MIMO system can transmit different signal waveforms. When continuous band is divided into several pieces, sparse frequency waveforms play an important role due to special spectrum. In this paper, we start from covariance matrix transmitted extend concept study pattern. With idea mind, first solve problem semidefinite...
According to the Tag application with function of covert communication, a method for sparse frequency waveform design based on radar-embedded communication is proposed. Firstly, waveforms are designed power spectral density fitting and quasi-Newton method. Secondly, eigenvalue decomposition sequence used get dominant space. Finally through projection orthogonal pseudorandom vectors in vertical subspace. Compared linear modulation waveform, can further improve bandwidth occupation signals,...
Sparse frequency waveform (SFW) is able to utilize spectrums from disjoint channels in a single so as improve the overall bandwidth that desirable both radar and communications. One challenge faced with sparse design relatively high range sidelobe level due spectrum discontinuity. In this paper, new approach suppression proposed exploits cancellation property of complementary codes (CC). Different criterions are facilitate requirements different applications. Numerical simulations have been...
Unmanned aerial vehicle (UAV) remote sensing images used for semantic segmentation possess distinct features compared to urban street scene images, including high resolution and a complex background. Spatial information plays pivotal role in enhancing the performance of high-resolution images. The dual-branch architecture incorporates supplementary branches capture spatial information. However, prior research on neglected interaction between contextual branches, leading suboptimal model...
Whether the down-tilt angle of mobile communication base station antenna is reasonable directly affects coverage effect and quality whole network. Measurement traditional parameters mainly relies on engineers climbing for manual measurements, confronting dilemma low measurement efficiency a dangerous working environment. Moreover, due to interference wind field in real unmanned aerial vehicle (UAV) scenes, there attitude disturbance during observation, which leads sharp increase error. To...
Datasets usually suffer from supervised information missing and weak generalization ability in deep convolution neural network. In this paper, pseudolabel (PL) of Weakly Supervised Learning (WSL) was used to address the problem missing, while Cross Network (CN) Multitask (MTL) solve PL, data predicted; thus, PL corresponding generated. CN, labeled were taken as two tasks train together. Firstly, divided into training dataset testing dataset, respectively, image preprocessing carried out....
This paper presents the design and realization of a large capacity airborne waveform recorder for UWB (ultra wide band) system based on FPGA Flash array. The has 80 MT29F64G08AJABA NAND Memory chips provides total 640GB storage capacity. With ADC working at 100MHz, time can be more than 100 minutes. For offline data analysis, Gigabit interface to PC, rate reach 25MB/s. device also applied in other measurement instruments systems. structure, implementation test results are described detail.
Synthetic Aperture Radar (SAR) has the all-day, allweather ability to work.It become one of important sensors for military reconnaissance and civilian remote sensing.Radio frequency interference (RFI) is major source low band SAR systems.This paper proposes a novel RFI suppression method Compressive Sensing SAR.To compressed sampled echo data, greedy algorithm adopted estimate spectrum with sparse feature, minimum description length (MDL) criteria used components sparsity.Then signal each...
Abstract Facial Beauty Prediction (FBP) is an important and challenging problem in the field of computer vision machine learning. Not only it easily prone to over-fitting due lack large-scale effective data, but also difficult quickly build robust face beauty evaluation models because variability facial appearance complexity human perception. Transfer learning can be able reduce dependence on large amounts data as well avoid overfitting problems. Broad Learning System (BLS) capable...