Abstract Herpes zoster (HZ) is an infectious dermatosis with high incidence worldwide. Age a key risk factor for HZ, and postherpetic neuralgia (PHN) the main sequelae. Until now, no index has been available to predict pathogenesis of PHN, rare reports have focused on immune response during aging PHN. In this study, we selected immunoglobulin complement proteins as markers humoral immunity, while T lymphocyte subsets natural killer (NK) cells were cell systematically study characteristics...

Due to the rich content of video data and excessive parameters 3D convolutional networks, how reduce difficulty parameter tuning training time is a difficult problem in action recognition. To solve these difficulties, new recognition algorithm based on dual-stream architecture improved three-dimensional aggregation proposed. The network structure convolution kernel split into two types kernels, spatial flow flow, fused network, residual introduced for optimization. verify effectiveness...

As an effective network protection method, computer user behavior anomaly detection can detect unknown attack behaviors. In order to anomalies more efficiently, the authors propose a model based on K-means algorithm. According actual characteristics of single-user behavior, algorithm uses sliding time window define transactions and first location strategy mine patterns. On this basis, fault-tolerant mode is adopted compare current with normal mode, results are obtained. Experiments show that...

This paper compares the effects of SE-Net and improved CBAM attention mechanism, proposes a CTSA (Channel Temporal Spatial Attention) mechanism that adds temporal for features, further TSA (Time-domain focuses on time-domain basis comparing multiple mechanisms. The experimental results show in field behaviour recognition, designed this can achieve better recognition when using lighter weight network structure.

The paper provides the method of improved genetic neural network for image segmentation. uses algorithm BP weights and thresholds to optimize, use definition bipolar fitness function mapping compression speed up training speed, then iterative achieve results experimental show that can better segmentation, compared with traditional method; Compared is greatly improved.

In order to improve the de-noising effect of wavelet threshold further, after analyzing shortcomings traditional soft and hard methods, this paper proposes a trade-off method. The typical blocks signal superposed standard Gaussian white noise in Matlab toolbox is taken as an example for simulating testing, verify feasibility availability new Signal-to-noise ratio (SNR) root-mean-square error (RMSE) are appraisal parameters performance. experiment results indicate superiority method, it...

Abstract In view of the characteristics low speed mobile ship network, strong network dynamics and large signal interference, this paper proposes a low-rate sensor routing algorithm. wireless networks, topology changes frequently is new challenge for resource constrained node. Constantly updated topology, nodes need to exchange data in order maintain fresh route, resulting consume lot energy. To solve problem, reduce control overhead considered be an effctive strategy. paper, we present...

This1 article introduces Software Defined Network (SDN) technology into the Wireless Sensor (WSN) field, working mechanism of OPENFLOW, and software defined wireless sensor network (SDN-WISE). The architecture is improved, data plane control WSN are decoupled. In-depth research on processing forwarding in intensified. Based SDN-WISE architecture, a rule-action strategy derived from OPENFLOW WSN. table stored at starting point communication link. route nodes do not need to check forward data,...

<p>Building information modelling (BIM) as an AEC digital approach is already far beyond its original definition; moreover, BIM has turned into a general project solution by providing single unified environment which able to integrate all sorts of entire and use them perform simulations analysis.</p><p>In one our complex high-rise office projects, BIM-integrated structural design was introduced in the early concept scheme stages modelling, analysis, optimization...

In order to effectively supervise the wearing of safety helmets by construction personnel, YOLOv4-tiny target detection algorithm is used detect helmets. A lightweight model with higher accuracy and less computation designed for YOLOv4-tiny, which more suitable real-time helmet detection. Firstly, G-Resblock replace Resblock reduce computational complexity occupy computing resources. However, prone error or missed in complex work scenarios. solve this problem, an attention mechanism added...

Forest fire is a destructive natural disaster, which characterized by uncontrollability and instantaneous explosiveness, making the prevention control of especially critical. In recent years, forest detection algorithms for automatic smoke recognition have received extensive attention from researchers at home abroad. However, in budding stage fire, target features are small usually hidden, makes it difficult traditional methods to achieve effective detection. Convolutional neural...

This paper studied GPS location information of the collection, processing, GPRS data transmission network, intelligent building security system features.Based on this techniques, we designed an embedded GPS/GPRS system, which is composed ARM9 microprocessor S3C2410, module and some peripheral devices.The running under Linux OS.The characteristics are flexible, powerful, high practicality.

The paper introduces current character recognition and using artificial neural network for pattern theory, studies the BP technique in system, expounds application of system basic principle, model analysis results recognition, some Suggestions are put forward.

when the amount of mobile Internet data is exploding, users need to be able enjoy high-quality services at any location and time. With characteristics massive high-density connections, it urgent study new network architecture optimization theory. New cloud wireless access theory research based on software-defined network.Decoupling plane, control plane management plane. Data responsible for transmission related operations, such as forwarding; Control call control, routing error flow etc.;...

The contamination caused by noise to the image leads missing key information or reflect role of real situation. Wavelet denoising based on spatial domain has a certain effect noise. When doing wavelet decomposition image, because function feature flexibility in base selection, different choices basis functions have effects noisy image. Simulation experiments are conducted Python for several commonly used using control variables method. final experimental results evaluated PSNR (Peak Signal...

Abstract Target tracking is widely used in video image, intelligent transportation, behavior analysis and other fields. At the same time, target technology faces more challenges. In this paper, occlusion scene studied, method based on spatiotemporal context combined with depth network model are introduced respectively. The main ideas principles of two methods analyzed, improvements these summarized. Finally, problems existing future development occluded algorithm prospected.

Event detection system is more and used in road monitoring. This paper proposes a traffic multi state recognition based on intelligent vision technology. The gray change interval of frame difference to update the image background, pixel control rule introduced as core software algorithm. Combined with monitoring data source, events such congestion, pedestrian parking are detected. actual test shows that can timely accurately detect related events, has high accuracy. As an important part...