A5101829389- Remote-Sensing Image Classification
- Remote Sensing and Land Use
- Face and Expression Recognition
- Advanced Image Fusion Techniques
- Advanced Image and Video Retrieval Techniques
- Image Retrieval and Classification Techniques
- Biometric Identification and Security
- Radiomics and Machine Learning in Medical Imaging
- Video Surveillance and Tracking Methods
- Lung Cancer Diagnosis and Treatment
- AI in cancer detection
- Infrared Target Detection Methodologies
- Face recognition and analysis
- Remote Sensing in Agriculture
- COVID-19 diagnosis using AI
- Cancer, Hypoxia, and Metabolism
- Digital Imaging for Blood Diseases
- Receptor Mechanisms and Signaling
- Neuroscience and Neuropharmacology Research
- Immune Response and Inflammation
- Energy Load and Power Forecasting
- Gait Recognition and Analysis
- Neuropeptides and Animal Physiology
- Adversarial Robustness in Machine Learning
- Spectroscopy and Chemometric Analyses
Southern University of Science and Technology
2025
Chongqing University of Technology
2014-2024
Chongqing University
2014-2024
Wenzhou Medical University
2017-2024
Affiliated Hospital of Youjiang Medical University for Nationalities
2024
State Key Laboratory of Medicinal Chemical Biology
2024
Tianjin International Joint Academy of Biomedicine
2024
Nankai University
2024
Southeast University
2024
Chongqing Medical University
2008-2023
The graph embedding (GE) methods have been widely applied for dimensionality reduction of hyperspectral imagery (HSI). However, a major challenge GE is how to choose the proper neighbors construction and explore spatial information HSI data. In this paper, we proposed an unsupervised algorithm called spatial-spectral manifold reconstruction preserving (SSMRPE) classification. At first, weighted mean filter (WMF) employed preprocess image, which aims reduce influence background noise....
Scene classification of high spatial resolution (HSR) images can provide data support for many practical applications, such as land planning and utilization, it has been a crucial research topic in the remote sensing (RS) community. Recently, deep learning methods driven by massive show impressive ability feature field HSR scene classification, especially convolutional neural networks (CNNs). Although traditional CNNs achieve good results, is difficult them to effectively capture potential...
Scene classification is an indispensable part of remote sensing image interpretation, and various convolutional neural network (CNN)-based methods have been explored to improve accuracy. Although they shown good performance on high-resolution (HRRS) images, discriminative ability extracted features still limited. In this letter, a high-performance joint framework combined CNNs vision transformer (ViT) (CTNet) proposed further boost the for HRRS scene classification. The CTNet method contains...
Scene classification is an active research topic in the remote sensing community, and complex spatial layouts with various types of objects bring huge challenges to classification. Convolutional neural network (CNN)-based methods attempt explore global features by gradually expanding receptive field, while long-range contextual information ignored. Vision transformer (ViT) can extract features, but learning ability local limited, it has a large computational complexity simultaneously. In...
Marginal Fisher analysis (MFA) exploits the margin criterion to compact intraclass data and separate interclass data, it is very useful analyze high-dimensional data. However, MFA just considers structure relationships of neighbor points, cannot effectively represent intrinsic hyperspectral imagery (HSI) that possesses many homogenous areas. In this paper, we propose a new dimensionality reduction (DR) method, termed local geometric (LGSFA), for HSI classification. Firstly, LGSFA uses points...
The graph embedding (GE) framework is very useful to extract the discriminative features of hyperspectral images (HSIs) for classification. However, a major challenge GE how select proper neighborhood size construction. To overcome this drawback, new semisupervised learning algorithm, which called sparse manifold analysis (S <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sup> MDA) method, was proposed by using manifold-based representation...
The scene classification of high spatial resolution (HSR) images is a challenging task in the remote sensing community. How to construct discriminative representation HSR key step improve performance. In this letter, we propose novel feature extraction method termed multilayer fusion network (MF <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> Net) for classification. At first, transferred VGGNet-16 model employed as extractor acquire...
Recently, the sparse representation (SR)-based graph embedding method has been extensively used in feature extraction (FE) tasks, but it is hard to reveal complex manifold structure and multivariate relationship of samples hyperspectral image (HSI). Meanwhile, small size sample problem HSI data also limits performance traditional SR approach. To tackle this problem, article develops a new semisupervised FE algorithm called geodesic-based hypergraph (GSMH). The presented first utilizes...
The spatial heterogeneity is an important indicator of the malignancy lung nodules in cancer diagnosis. Compared with 2D nodule CT images, 3D volumes entire objects hold richer discriminative information. However, for deep learning methods driven by massive data, effectively capturing features limited labeled samples a challenging task. Different from previous models that proposed transfer pattern or scratch models, we develop self-supervised based on domain adaptation (SSTL-DA) CNN...
Hyperspectral images (HSIs) possess a large number of spectral bands, which easily lead to the curse dimensionality. To improve classification performance, huge challenge is how reduce bands and preserve valuable intrinsic information in HSI. In this letter, we propose novel unsupervised dimensionality reduction method called local neighborhood structure preserving embedding (LNSPE) for HSI classification. At first, LNSPE reconstructs each sample with its neighbors obtains optimal weights...
High spatial resolution remote sensing (HSRRS) images contain complex geometrical structures and patterns, thus HSRRS scene classification has become a significant challenge in the community. In recent years, convolutional neural network (CNN)-based methods have attracted tremendous attention obtained excellent performance classification. However, traditional CNN-based focus on processing original red-green-blue (RGB) image-based features or single-layer to achieve representation, ignore...
Sparse representation-based graph embedding methods have been successfully applied to dimensionality reduction (DR) in recent years. However, these approaches usually become problematic the presence of hyperspectral image (HSI) that contains complex nonlinear manifold structure. Inspired by progress learning and hypergraph framework, a novel DR method named local constraint-based sparse (LC-SMHL) algorithm is proposed discover manifold-based structure multivariate discriminant relationship...
Scene classification of high-resolution images is an active research topic in the remote sensing community. Although convolutional neural network (CNN)-based methods have obtained good performance, large-scale changes ground objects complex scenes restrict further improvement accuracy. In this letter, a global–local dual-branch structure (GLDBS) designed to explore discriminative features original and crucial areas, strategy decision-level fusion applied for performance improvement. To...
Deep belief networks (DBNs) have been widely applied in hyperspectral imagery (HSI) processing. However, the original DBN model fails to explore prior knowledge of training samples which limits discriminant capability extracted features for classification. In this paper, we proposed a new deep learning method, termed manifold-based multi-DBN (MMDBN), obtain manifold HSI. MMDBN designed hierarchical initialization method that initializes network by local geometric structure hidden data. On...
The main challenge of scene classification is to understand the semantic context information high-resolution remote sensing images. Although vision transformer (ViT)-based methods have been explored boost long-range dependencies images, connectivity between neighboring windows still limited. Meanwhile, ViT-based commonly contain a large number parameters, resulting in huge computational consumption. In this paper, novel lightweight dual-branch swin (LDBST) method for proposed, and...
A large number of coverage criteria to generate tests from logical expressions have been proposed. Although there variations in the terminology, articulation and original source expressions, many these are fundamentally same. The most commonly known widely used criterion is that modified condition decision (MCDC), but some articulations MCDC had ambiguities. This has led confusion on part testers, students, tool developers how best implement test criteria. paper presents a complete...
Hyperspectral image (HSI) generally contains a complex manifold structure and strong sparse correlation in its nonlinear high-dimensional data space. However, the existing learning methods usually consider relationship separately rather than combining properties to discover intrinsic information original data. To simultaneously reveal relation of HSI, novel feature extraction (FE) method, called local manifold-based discriminant (LMSDL), has been proposed on basis representation (SR). The...
Pine wilt disease (PWD), caused by the pine wood nematode (Bursaphelenchus xylophilus), is a global destructive threat to forests and has led serious economic losses all over world. Therefore, it necessary establish feasible effective method accurately monitor estimate PWD infection. In this study, we used hyperspectral imagery (HI) collected an unmanned airship with imaging spectrometer detect in healthy, early, middle infection stages. To avoid massive calculations on full spectral...
Building extraction is a critical part of remote sensing (RS) image interpretation, and it popular research topic in the RS community. However, building from images difficult task due to its various shape, size, complex scene. The extracted feature existing deep learning methods lack discrimination, resulting incomplete buildings irregular boundaries. Most studies are mainly concentrated on urban areas, ignoring illegal blue-roofed rural areas. To address above-mentioned problems,...
Currently, the integration of Global Navigation Satellite System (GNSS), Ultra-Wideband (UWB), and Inertial (INS) has become a reliable positioning method for outdoor dynamic vehicular airborne applications, enabling high-precision continuous in complex environments. However, environmental interference limitations single sources pose challenges. Especially areas with limited access to satellites UWB base stations, loosely coupled frameworks GNSS/INS UWB/INS are insufficient support robust...