A5103247632- Remote-Sensing Image Classification
- Advanced Image Fusion Techniques
- Remote Sensing and Land Use
- Robotic Path Planning Algorithms
- Image Retrieval and Classification Techniques
- Geochemistry and Geologic Mapping
- Advanced Graph Neural Networks
- Spectroscopy and Chemometric Analyses
- Optimization and Search Problems
- Anomaly Detection Techniques and Applications
- Robot Manipulation and Learning
- Target Tracking and Data Fusion in Sensor Networks
- Advanced Measurement and Detection Methods
- AI-based Problem Solving and Planning
- Complex Network Analysis Techniques
- Advanced Chemical Sensor Technologies
- Distributed Sensor Networks and Detection Algorithms
- Auction Theory and Applications
- Advanced Image and Video Retrieval Techniques
Suzhou University of Science and Technology
2021-2024
University of Science and Technology of China
2021-2024
Hebei Normal University
2021-2022
Beijing Normal University
2019-2022
Kunming University of Science and Technology
2019
Convolutional neural networks (CNNs) have attained remarkable performance in hyperspectral image (HSI) classification. However, the existing CNNs are restricted by their limited receptive field HSI Recently, transformer proved to be promising many tasks thanks global field, but they easily ignore some local information that is important for In this letter, we propose a novel method entitled convolutional network (CTN) order make full use of spectral and spatial information, adopts center...
Classification is one of the most important research topics in hyperspectral image (HSI) analyses and applications. Although convolutional neural networks (CNNs) have been widely introduced into study HSI classification with appreciable performance, misclassification problem pixels on boundary adjacent land covers still significant due to interfering neighboring whose categories are different from target pixel. To address this challenge, article, we propose a center attention network for...
Hyperspectral image (HSI) classification with a small number of training samples has been an urgently demanded task because collecting labeled for hyperspectral data is expensive and time-consuming. Recently, graph attention network (GAT) shown promising performance by means semisupervised learning. It combines the information unlabeled so that weakness inadequate alleviated. In this letter, we propose novel method, spectral–spatial GAT (SSGAT), HSI classification. The proposed SSGAT takes...
Classifying Hyperspectral images with few training samples is a challenging problem. The generative adversarial networks (GAN) are promising techniques to address the problems. GAN constructs an game between discriminator and generator. generator generates that not distinguishable by discriminator, determines whether or sample composed of real data. In this paper, introducing multilayer features fusion in dynamic neighborhood voting mechanism, novel algorithm for HSIs classification based on...
Hyperspectral unmixing is an important task in the analyses and applications of hyperspectral images. Recently, autoencoder network has been intensively studied to unmix image, recovering material signatures their corresponding abundance maps from pixels. However, cannot get a unique solution since loss function nonconvex. In addition, data often contain lot noise. To address these problems, we propose network, referred as MDC-SAE, that introduces two different constraints optimize spectral...
Hyperspectral data contains abundant spectral do main information, which is of great significance to classification objects. However, due the lack labeled data, it difficult get an acceptable result by just using small number data. We propose a semi-supervised model based on convolutional neural network and introduce attention mechanism balance sample weight. After convolution multi-layer network, more information concentrated channels, so we use Squeeze-and-Excitation block, can adaptively...
The era of Industry 4.0 is gradually transforming our society into a data-driven one, which can help us uncover valuable information from accumulated data, thereby improving the level social governance. detection anomalies, crucial for maintaining societal trust and fairness, yet it poses significant challenges due to ubiquity anomalies difficulty in identifying them accurately. This paper aims enhance performance current Graph Convolutional Network (GCN)-based Anomaly Detection (GAD)...
Hyperspectral images (HSIs) contain a large number of mixed pixels due to low spatial resolution, which poses great challenges the analyses and applications HSIs. In recent years, convolutional neural networks (CNNs) have attained promising performance in HSI field. However, few CNN-based methods are proposed solve hyperspectral unmixing (HU) problem because insufficient labeled samples. this paper, we propose novel unsupervised method, sparsity constrained autoencoder network (SC-CAE), for...
In recent years, the rapid development of robots has brought many conveniences to production and life human beings. The research on is great significance. Task allocation one most important issues in field multi-robot research. As working environment inherently highly dynamic uncertain, multi-robots also have problem communication cooperation, how quickly accurately allocate tasks these complex environments very important. this paper, aiming at real-time task assignment agreed time coupling,...
Aiming at the difficulty of noise estimation and outlier in data fusion, a fuzzy theory fusion algorithm based on confidence distance is proposed its application sensor monitoring studied. Firstly, according to characteristics values, traditional methods are analyzed. Secondly, used describe proximity between data, erroneous eliminated. Finally, merged by get result fusion. Experiments show that difference value small, close actual state measured object.
Convolutional neural network has been widely used in hyperspectral image classification. Compared with the early machine learning method, it made great progress. However, convolution kernel classification ignores intrinsic relationship among spatial pixels when extracting spectral features, which will lead to poor contour and very small false prediction results. Besides, The data can only be labeled by experts, requires a lot of labor material resources. In order improve accuracy images...
In recent years, the service robot has been developed rapidly, which brought a lot of convenience for industries and human life. It is great significance to study robot. Mission planning key component robot, used solve sequential decision problem Because highly dynamic uncertain working environment how model deal with these complex environments very important agents complete mission correctly efficiently. this paper, multimission method based on L-HMM utility function description considering...