A5046467104- Face and Expression Recognition
- Remote-Sensing Image Classification
- Spectroscopy and Chemometric Analyses
- Image Retrieval and Classification Techniques
- Anomaly Detection Techniques and Applications
- Advanced Image Processing Techniques
- Advanced Image Fusion Techniques
- Image and Signal Denoising Methods
- Medical Imaging and Analysis
- Remote Sensing and Land Use
- Video Surveillance and Tracking Methods
- Flood Risk Assessment and Management
- Data Stream Mining Techniques
- Network Security and Intrusion Detection
- Gene expression and cancer classification
- Remote Sensing and LiDAR Applications
- Geological Modeling and Analysis
- Computer Graphics and Visualization Techniques
- Radiation Dose and Imaging
- Medical Imaging Techniques and Applications
- Sparse and Compressive Sensing Techniques
- Neural Networks and Applications
- Machine Learning and ELM
- Simulation and Modeling Applications
- Hydrology and Sediment Transport Processes
Zhejiang Wanli University
2022-2024
Zhejiang University of Technology
2012-2018
Online anomaly detection is a key challenge for industrial internet of things (IIoT) applications, as anomalies may occur in data streams from sensors and cause losses or damages. However, most existing methods online have limitations efficiency, effectiveness timeliness, especially with the massive distributed IIoT devices. Therefore, developing stream processing framework to discover time ensure proper operation system an urgent issue IIoT. In this paper, we propose flexible that enables...
Low-Dose computer tomography (LDCT) is an ideal alternative to reduce radiation risk in clinical applications. Although supervised-deep-learning-based reconstruction methods have demonstrated superior performance compared conventional model-driven algorithms, they require collecting massive pairs of low-dose and norm-dose CT images for neural network training, which limits their practical application LDCT imaging. In this paper, we propose unsupervised training data-free learning method...
Feature is important for many applications in biomedical signal analysis and living system analysis. A fast discriminative stochastic neighbor embedding (FDSNE) method feature extraction proposed this paper by improving the existing DSNE method. The algorithm adopts an alternative probability distribution model constructed based on its<mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M1"><mml:mrow><mml:mi>K</mml:mi></mml:mrow></mml:math>-nearest neighbors from interclass...
Aiming at the problems of missing important features, inconspicuous details and unclear textures in fusion multimodal medical images, this paper proposes a method computed tomography (CT) image magnetic resonance imaging (MRI) using generative adversarial network (GAN) convolutional neural (CNN) under enhancement. The generator aimed high-frequency feature images used double discriminators to target after inverse transform; Then were fused by trained GAN model, low-frequency CNN pre-training...
This article aims to combine deep learning with image analysis technology and propose an effective classification method for distal radius fracture types. Firstly, extended U-Net three-layer cascaded segmentation network was used accurately segment the most important joint surface non areas identifying fractures. Then, images of area separately were classified trained distinguish Finally, based on results two images, normal or ABC could be comprehensively determined. The accuracy rates...
The accurate extraction of surface rivers is great significance to ecology, residence and so on. In view the incomplete recognition river edge contour in remote sensing image classical deep learning network U-Net, ability learn retain detailed information feature map enhanced by strengthening attention mechanism introducing densely connected Atrous Spatial Pyramid Pooling on basis U-Net. experimental results show that Pixel Accuracy water this method 92.1%, Mean Intersection Over Union up...
Big data has the traits such as “the curse of dimensionality,” high storage cost, and heavy computation burden. Self-representation-based feature extraction methods cannot effectively deal with image-level structural noise in data, so how to character a better relationship reconstruction representation is very important. Recently, sparse smoothed matrix multivariate elliptical distribution (SMED) using information handle low-rank error images caused by illumination or occlusion been...
A new method for performing a nonlinear form of manifold-oriented stochastic neighbor projection is proposed. By the use kernel functions, one can operate in feature space without ever computing coordinates data that space, but rather by simply inner products between images all pairs space. The proposed termed as kernel-based manifoldoriented projection(KMSNP). two different strategies, KMSNP divided into methods: KMSNP1 and KMSNP2. Experimental results on several databases show that,...
Change detection is an important branch in remote sensing image processing. Deep learning has been widely used this field. In particular, a wide variety of attention mechanisms have made great achievements. However, some models become increasingly complex and large, often unfeasible for edge applications. This poses major obstacle to industrial paper, solve the above challenges, we propose Lightweight network structure improve results while taking into account efficiency. Specifically,...
With high spectral resolution, hyperspectral image(HSI) data will result in the Hughes phenomenon, which brings a huge challenge to image classification(HIC). Feature extraction can be applied address this problem. But several traditional methods often ignore spatial structure information of HSI data. In paper, we propose tensor nuclear norm based matrix regression projections(TNMRP) for feature images. Firstly, TNMRP preprocesses by filling method. Then, it automatically builds graph...
By improving on the Quad-tree LOD algorithm, this paper proposes a new algorithm based terrain partitioning and simplifying by using grid models of different levels detail dynamically.This can be reduced number triangular facets in scene.The experimental results show that generate continuous multi-resolution model improve efficiency real-time rendering.