A5084697576- Non-Invasive Vital Sign Monitoring
- Advanced SAR Imaging Techniques
- Indoor and Outdoor Localization Technologies
- Gait Recognition and Analysis
- Context-Aware Activity Recognition Systems
- Structural Health Monitoring Techniques
- Advanced Measurement and Detection Methods
- Wireless Networks and Protocols
- Machine Fault Diagnosis Techniques
- Microwave Imaging and Scattering Analysis
- Radar Systems and Signal Processing
- Anomaly Detection Techniques and Applications
University of Electronic Science and Technology of China
2017-2024
Beijing Health Vocational College
2024
CRRC (China)
2021
The accurate classification of activity patterns based on radar signatures is still an open problem and a key to detect anomalous behavior for security health applications. This paper presents novel iterative convolutional neural network strategy with autocorrelation pre-processing instead the traditional micro-Doppler image classify activities or subjects accurately. proposed uses deep learning framework automatic definition extraction features. followed by supervised classifier label...
Classification of different human activities using micro-Doppler data and features is considered in this study, focusing on the distinction between walking running. 240 recordings from 2 subjects were collected a series simulations performed real motion Carnegie Mellon University Motion Capture Database. The maximum frequency shift period duration are utilized as two classification criterions. Numerical results compared against several techniques including Linear Discriminant Analysis (LDA),...
Radar micro-Doppler signatures have been proposed for human monitoring and activity classification surveillance outdoor security, as well ambient assisted living in healthcare-related applications. A known issue is the performance reduction when target moving tangentially to line of sight radar. Multiple techniques address this, such multistatic radar some extent, interferometric (IF) simulator presented generate synthetic data representative eight systems (monostatic, circular in-line [IM]...
Falls in the elderly represent a serious challenge for global population. To address it, monitoring of daily living has been suggested, with radar emerging to be useful platform it due its various benefits acceptance and privacy. Here, we show results from use an S band activity detection importance selecting specific frequency bins improve suitability human movement classification. The feature selection key activities such as falls presented. Initial 65% are improved 85% further 90%...
The accurate classification of WiFi-based activity patterns is still an open problem and critical to detect behavior for non-visualization applications. This paper proposes a novel approach that uses IQ data short-time Fourier transform (STFT) time-frequency images automatically accurately classify human activities. offsets features, calculated from time-domain values one-dimensional principal component analysis (1D-PCA) two-dimensional (2D-PCA) values, are applied as features input the...
This study introduces an innovative approach by incorporating statistical offset features, range profiles, time–frequency analyses, and azimuth–range–time characteristics to effectively identify various human daily activities. Our technique utilizes nine feature vectors consisting of six features three principal component analysis network (PCANet) fusion attributes. These are derived from combined elevation azimuth data, considering their spatial angle relationships. The attributes generated...
This paper presents a novel approach that applies WiFi-based IQ data and time–frequency images to classify human activities automatically accurately. The proposed strategy first uses the Choi–Williams distribution transform Margenau–Hill spectrogram obtain images, followed by offset principal component analysis (PCA) feature extraction. features were extracted from several spectra with maximum energy values in time domain, PCA via whole image slices on them rich unit information. Finally,...
This paper presents a general approach for extracting the specific feature directly from all Intrinsic Mode Functions (IMF). It is believed that every IMF possessed with its own physical meaning. However, detailed signature part of specially determined IMF. Due to various numbers as well frequency mean deviations, it costs much time choose certain IMF, which also easy make mistakes. In this study, in order prevent making mistake novel algorithm put forward on basis analysis between adjacent...