A5101896142- Remote-Sensing Image Classification
- Advanced Image Fusion Techniques
- Anomaly Detection Techniques and Applications
- Spectroscopy and Chemometric Analyses
- Sparse and Compressive Sensing Techniques
- Remote Sensing and Land Use
- Water Quality Monitoring and Analysis
- Advanced Chemical Sensor Technologies
- Nanoplatforms for cancer theranostics
- Automated Road and Building Extraction
- Power System Optimization and Stability
- Photodynamic Therapy Research Studies
- Coal and Coke Industries Research
- Image and Signal Denoising Methods
- Geochemistry and Geologic Mapping
- Tensor decomposition and applications
- Image Processing and 3D Reconstruction
- Coal and Its By-products
- Photoacoustic and Ultrasonic Imaging
- Image Retrieval and Classification Techniques
- Water Quality Monitoring Technologies
- Seismic Imaging and Inversion Techniques
- Minerals Flotation and Separation Techniques
- Energy Load and Power Forecasting
- Time Series Analysis and Forecasting
Rice University
2025
University of Tennessee at Knoxville
2013-2023
Tianjin Economic-Technological Development Area
2022
China University of Petroleum, East China
2019
North Minzu University
2017
Knoxville College
2015
Anomaly detection has been known to be a challenging problem due the uncertainty of anomaly and interference noise. In this paper, we focus on in hyperspectral images (HSIs) propose novel algorithm based spectral unmixing dictionary-based low-rank decomposition. The innovation is threefold. First, highly mixed nature pixels HSI data, instead using raw pixel directly for detection, proposed applies obtain abundance vectors uses these detection. We show that possess more distinctive features...
Hyperspectral image unmixing is the process of estimating pure source signals (endmemebers) and their proportions (abundances) from highly mixed spectroscopic images. Due to model inaccuracies observation noise, has been a very challenging problem. In this paper, we exploit potential using autoencoder tackle challenges. Two important facts are considered in algorithm: first, noise hyperspectral generally exists largely affects results; second, mixing contains sparsity priori which should be...
Anomaly detection becomes increasingly important in hyper-spectral image analysis, since it can now uncover many material substances which were previously unresolved by multi-spectral sensors. In this paper, we propose a Low-rank Tensor Decomposition based anomaly Detection (LTDD) algorithm for Hyperspectral Imagery. The HSI data cube is first modeled as dense low-rank tensor plus sparse tensor. Based on the obtained tensor, LTDD further decomposes using Tucker decomposition to extract core...
A situational awareness system is essential to provide accurate understanding of power dynamics, such that proper actions can be taken in real time response disturbances and avoid cascading blackouts. Event analysis has been an important component any system. However, most state-of-the-art techniques only handle single event analysis. This paper tackles the challenging problem multiple detection recognition. We propose a new conceptual framework, referred as unmixing, where we consider...
Anomaly detection has been known to be a challenging, ill-posed problem due the uncertainty of anomaly and interference noise. In this paper, we propose novel low rank algorithm in hyperspectral images (HSI), where three components are involved. First, highly mixed nature pixels HSI, instead using raw pixel directly for detection, proposed applies spectral unmixing algorithms obtain abundance vectors uses these detection. Second, better classification, dictionary is built based on mean-shift...
Traditional hyperspectral anomaly detection methods either model the global background or local neighborhood, that bring some apparent drawbacks, such as unreasonable assumption of uni-modular in detectors, high false alarms by sliding windows detectors. In this paper, a source component-based approach is proposed. It first extracts components spectral image data cube using blind component separation and then identifies are (or salient) to other components. We interpret matrix decomposition...
The ChemCam instrument package on the Mars rover, "Curiosity", is first planetary that employs laser-induced breakdown spectroscopy (LIBS) to determine compositions of geological samples another planet. However, sampled spectra are often corrupted by various sources interferences would largely affect accuracy elemental concentration estimation. Therefore, preprocessing essential improve quality spectra. This paper revisits conventional procedures where denoising followed continuum removal....
The effective integration of aerial remote sensing data and ground multi-source has always been one the difficulties quantitative sensing. A new monitoring mode is designed, which installs hyperspectral imager on UAV places a buoy spectrometer river. Water samples are collected simultaneously to obtain in situ assay total phosphorus, nitrogen, COD, turbidity, chlorophyll during collection. cross-correlogram spectral matching (CCSM) algorithm used match with significantly reduce noise. An...
Unsupervised spectral unmixing (i.e., endmember extraction and abundance estimation) of nonlinear mixture is a very challenging subject in hyperspectral image analysis. In this paper, we present new interpretation the reflectance by normalizing absolute value into unit L <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sub> norm vector, such that reading can be treated as probability distribution. The then interpreted possibility distribution...
Classification of the hyperspectral imagery highly rely on selected samples. Learning from samples based their respective distributions other than treating them identical is more reasonable. Based law universal gravitation, a novel classification method proposed in this paper. In method, category label pixel determined by comparing gravitational force between different classes. A larger class means likely belongs to particular class. Moreover, log function introduced calculation gravitation...
This work presents a novel deep learning model for early, accurate, and robust detection, recognition, temporal localization of multi-type events in large-scale power systems. The proposed method develops unified 1-D fully convolutional network (FCN) that takes time series raw frequency signals measured from system as input, extracts distinguishing features, predicts at every point the if an event is happening what type is. Compared to existing methods, eliminates necessity hand-crafted...