The highly correlated spectral features and the limited training samples pose challenges in hyperspectral image classification. In this article, to tackle issues of end-to-end feature learning transfer with labeled samples, we propose a unified multiscale (UML) framework, which is based on fully convolutional network. A spatial-channel attention mechanism shuffle block are proposed UML framework improve problem land-cover map distortion. contextual information enhanced before last...

Recently, the deep learning algorithms have been increasingly utilized in remote sensing change detection. However, incomplete buildings and blurred edges caused by complex scenes detection applications make results fail to describe real land cover changes. Superpixels can be used alleviate edge blurring, but existing superpixel methods cannot trained jointly with models In this work, we investigated an innovative double-head method using learning, called double U-Net (W-Net), which consists...

Obtaining high-precision soil organic matter (SOM) spatial distribution information is of great significance for applications such as precision agriculture. But in the current hyperspectral SOM inversion work, moisture greatly influences representation sensitive on spectrum. Therefore, a Kubelka-Munk theory based spectral correction model removal proposed to improve sensitivity SOM. Firstly, content was obtained by use physical and an unmixing method. Then, implemented quantitative...

Ground surface subsidence is a universal phenomenon in coal mining areas which can cause serious damage to the surrounding environment. In this paper, we consider use of differential interferometric synthetic aperture radar (D-InSAR), multi-temporal InSAR (MT-InSAR), and pixel offset tracking technique monitor deformation area. study, two-pass D-InSAR method generate 19 image pairs from 20 TerraSAR-X SpotLight images. The results show that be used obtain high accuracy where there no gradient...

Practical Byzantine Fault Tolerance (PBFT) is a deterministic consensus algorithm, which widely used in blockchain due to the advantage of no forks. However, PBFT still has some problems, such as not being suitable for dynamic networks, requiring large communication overhead, and able identify nodes. Therefore, we propose Extensible-PBFT (EPBFT) it can take different steps reach according network environment system. Firstly, add election nodes based on verifiable random function (VRF), makes...

Soil organic matter (SOM) concentration is an important factor affecting soil quality, and rapid wide-scale monitoring of SOM a key step toward sustainable agriculture. Hyperspectral technology widely used in composition monitoring, due to its rich spectral information. The complex imaging environment hyperspectral imagery the mixed pixel problem have led current applications condition estimation mostly using data-driven methods. However, process based on methods cannot be explained by...

While many process mining algorithms have been proposed recently, there does not exist a widely accepted benchmark to evaluate and compare these algorithms. As result, it can be difficult choose suitable algorithm for given enterprise or application domain. Some recent systems developed address this issue. However, evaluating available against large set of business models (e.g., in enterprise) computationally expensive, tedious, time-consuming. This paper investigates scalable solution that...

An implicit assumption in many generic object trackers is that the videos are blur free. However, motion very common real videos. The performance of a tracker may drop significantly when it applied to with severe blur. In this paper, we propose new Tracking-Learning-Data approach transfer blur-invariant without deblurring image sequences. Before tracking, large set unlabeled images used learn objects' visual prior knowledge, which then transferred appearance model specific target. During...

It has been widely certified that hyperspectral images can be effectively used to monitor soil organic matter (SOM). Though numerous bands reveal more details in spectral features, information redundancy and noise interference also come accordingly. Due the fact that, nowadays, prevailing dimensionality reduction methods targeted fail make effective band selections, it is hard capture features of ground objects quickly accurately. In this paper, solve inefficiency instability feature...

Detecting gait events from video data accurately would be a challenging problem. However, most detection methods for are currently based on wearable sensors, which need high cooperation users and power consumption restriction. This study presents novel algorithm achieving accurate of toe-off using single 2D vision camera without the participants. First, set feature, namely consecutive silhouettes difference maps (CSD-maps), is proposed to represent pattern. A CSD-map can encode several...

With an increase in the volume of threats Cyber threat defense, knowledge graph is a good solution for cyber intelligence (CTI) to rapidly analyze advanced situations. Cybersecurity named entity recognition (Cs-NER) critical task identify cybersecurity related terms records and achieved significant success using deep neural network with word2vec method. This paper introduces BERT language representation model as pretrained Long Short-Term Memory (LSTM), Iterated Dilated Convolutional Neural...

In head mounted display (HMD), in order to cancel pincushion distortion, the images displayed on mobile should be pre-warped with barrel distortion. The copyright of video verified both original view and virtual view. A robust watermarking resistant against distortion for HMDs is proposed this paper. Watermark mask embedded into image consideration imperceptibility robustness watermarking. detect watermark from an estimation method HMDs. Then, same warp enforced estimated parameters...

The face recognition methods based on convolutional neural network have achieved great success. existing model usually used the residual as core architecture. is good at reusing features, but it difficult to explore new features. And densely connected can be We proposed a named Dense Face performance of in recognition. and composed Block layers, transition layers classification layer. was trained with joint supervision center loss softmax through feature normalization enabled learn more...