Attention-Based Residual Dense Shrinkage Network for ECG Denoising

Electrocardiogram(ECG)signal is one of the noninvasive physiological measurement techniques commonly usedin cardiac diagnosis.However,in real scenarios,the ECGsignal is susceptible to various noise erosion,which affectsthe subsequent pathological analysis.Therefore,the effective removal of the noise from ECG signals has becomea top priority in cardiac diagnostic research.Aiming at the problem of incomplete signal shape retention andlow signal-to-noise ratio(SNR)after denoising,a novel ECG denoising network,named attention-based residualdense shrinkage network(ARDSN),is proposed in this paper.Firstly,the shallow ECG characteristics are extractedby a shallow feature extraction network(SFEN).Then,the residual dense shrinkage attention block(RDSAB)isused for adaptive noise suppression.Finally,feature fusion representation(FFR)is performed on the hierarchicalfeatures extracted by a series of RDSABs to reconstruct the de-noised ECG signal.Experiments on the MIT-BIHarrhythmia database and MIT-BIH noise stress test database indicate that the proposed scheme can effectively resistthe interference of different sources of noise on the ECG signal.

Enhancing Dense Small Object Detection in UAV Images Based on Hybrid Transformer

Transformer-based models have facilitated significant advances in object detection.However,their extensive computational consumption and suboptimal detection of dense small objects curtail their applicability in unmanned aerial vehicle(UAV)imagery.Addressing these limitations,we propose a hybrid transformer-based detector,H-DETR,and enhance it for dense small objects,leading to an accurate and efficient model.Firstly,we introduce a hybrid transformer encoder,which integrates a convolutional neural network-based cross-scale fusion module with the original encoder to handle multi-scale feature sequences more efficiently.Furthermore,we propose two novel strategies to enhance detection performance without incurring additional inference computation.Query filter is designed to cope with the dense clustering inherent in drone-captured images by counteracting similar queries with a training-aware non-maximum suppression.Adversarial denoising learning is a novel enhancement method inspired by adversarial learning,which improves the detection of numerous small targets by counteracting the effects of artificial spatial and semantic noise.Extensive experiments on the VisDrone and UAVDT datasets substantiate the effectiveness of our approach,achieving a significant improvement in accuracy with a reduction in computational complexity.Our method achieves 31.9%and 21.1%AP on the VisDrone and UAVDT datasets,respectively,and has a faster inference speed,making it a competitive model in UAV image object detection.

MSADCN:Multi-Scale Attentional Densely Connected Network for Automated Bone Age Assessment

Bone age assessment(BAA)helps doctors determine how a child’s bones grow and develop in clinical medicine.Traditional BAA methods rely on clinician expertise,leading to time-consuming predictions and inaccurate results.Most deep learning-based BAA methods feed the extracted critical points of images into the network by providing additional annotations.This operation is costly and subjective.To address these problems,we propose a multi-scale attentional densely connected network(MSADCN)in this paper.MSADCN constructs a multi-scale dense connectivity mechanism,which can avoid overfitting,obtain the local features effectively and prevent gradient vanishing even in limited training data.First,MSADCN designs multi-scale structures in the densely connected network to extract fine-grained features at different scales.Then,coordinate attention is embedded to focus on critical features and automatically locate the regions of interest(ROI)without additional annotation.In addition,to improve the model’s generalization,transfer learning is applied to train the proposed MSADCN on the public dataset IMDB-WIKI,and the obtained pre-trained weights are loaded onto the Radiological Society of North America(RSNA)dataset.Finally,label distribution learning(LDL)and expectation regression techniques are introduced into our model to exploit the correlation between hand bone images of different ages,which can obtain stable age estimates.Extensive experiments confirm that our model can converge more efficiently and obtain a mean absolute error(MAE)of 4.64 months,outperforming some state-of-the-art BAA methods.

Shear wave splitting analysis of local earthquakes from dense arrays in Shimian,Sichuan

The Shimian area of Sichuan sits at the junction of the Bayan Har block.Sichuan-Yunnan rhombic block,and Yangtze block,where several faults intersect.This region features intense tectonic activity and frequent earthquakes.In this study,we used local seismic waveform data recorded using dense arrays deployed in the Shimian area to obtain the shear wave splitting parameters at 55 seismic stations and thereby determine the crustal anisotropic characteristics of the region.We then analyzed the crustal stress pattern and tectonic setting and explored their relationship in the study area.Although some stations returned a polarization direction of NNW-SSE.a dominant polarization direction of NW-SE was obtained for the fast shear wave at most seismic stations in the study area.The polarization directions of the fast shear wave were highly consistent throughout the study-area.This orientation was in accordance with the direction of the regional principal compressive stress and parallel to the trend of the Xianshuihe and Daliangshan faults.The distribution of crustal anisotropy in this area was affected by the regional tectonic stress field and the fault structures.The mean delay time between fast and slow shear waves was 3.83 ms/km.slightly greater than the values obtained in other regions of Sichuan.This indicates that the crustal media in our study area had a high anisotropic strength and also reveals the influence of tectonic complexity resulting from the intersection of multiple faults on the strength of seismic anisotropy.

Siamese Dense Pixel-Level Fusion Network for Real-Time UAV Tracking

Onboard visual object tracking in unmanned aerial vehicles(UAVs)has attractedmuch interest due to its versatility.Meanwhile,due to high precision,Siamese networks are becoming hot spots in visual object tracking.However,most Siamese trackers fail to balance the tracking accuracy and time within onboard limited computational resources of UAVs.To meet the tracking precision and real-time requirements,this paper proposes a Siamese dense pixel-level network for UAV object tracking named SiamDPL.Specifically,the Siamese network extracts features of the search region and the template region through a parameter-shared backbone network,then performs correlationmatching to obtain the candidate regionwith high similarity.To improve the matching effect of template and search features,this paper designs a dense pixel-level feature fusion module to enhance the matching ability by pixel-wise correlation and enrich the feature diversity by dense connection.An attention module composed of self-attention and channel attention is introduced to learn global context information and selectively emphasize the target feature region in the spatial and channel dimensions.In addition,a target localization module is designed to improve target location accuracy.Compared with other advanced trackers,experiments on two public benchmarks,which are UAV123@10fps and UAV20L fromthe unmanned air vehicle123(UAV123)dataset,show that SiamDPL can achieve superior performance and low complexity with a running speed of 100.1 fps on NVIDIA TITAN RTX.

Low Complexity Joint Spectrum Resource and Power Allocation for Ultra Dense Networks

In this paper,we propose a low complexity spectrum resource allocation scheme cross the access points(APs)for the ultra dense networks(UDNs),in which all the APs are divided into several AP groups(APGs)and the total bandwidth is divided into several narrow band spectrum resources and each spectrum resource is allocated to APGs independently to decrease the interference among the cells.Furthermore,we investigate the joint spectrum and power allocation problem in UDNs to maximize the overall throughput.The problem is formulated as a mixed-integer nonconvex optimization(MINCP)problem which is difficult to solve in general.The joint optimization problem is decomposed into two subproblems in terms of the spectrum allocation and power allocation respectively.For the spectrum allocation,we model it as a auction problem and a combinatorial auction approach is proposed to tackle it.In addition,the DC programming method is adopted to optimize the power allocation subproblem.To decrease the signaling and computational overhead,we propose a distributed algorithm based on the Lagrangian dual method.Simulation results illustrate that the proposed algorithm can effectively improve the system throughput.

Spectrum-Efficient User Association in Caching Enabled Dense Small Cell Network

In this paper,we reveal the fundamental limitation of network densification on the performance of caching enabled small cell network(CSCN)under two typical user association rules,namely,contentand distance-based rules.It indicates that immoderately caching content would significantly change the interference distribution in CSCN,which may degrade the network area spectral efficiency(ASE).Meanwhile,it is shown that content-based rule outperforms the distance-based rule in terms of network ASE only when small cell base stations(BSs)are sparsely deployed with low decoding thresholds.Moreover,it is proved that network ASE under distance-based user association serves as the upper bound of that under content-based rule in dense BS regime.To enable more spectrum-efficient user association in dense CSCN,we further optimize network ASE by designing a probabilistic content retrieving strategy based on distance-based rule.With the optimized retrieving probability,network ASE could be substantially enhanced and even increase with the growing BS density in dense BS regime.

Dense Spatial-Temporal Graph Convolutional Network Based on Lightweight OpenPose for Detecting Falls

Fall behavior is closely related to high mortality in the elderly,so fall detection becomes an important and urgent research area.However,the existing fall detection methods are difficult to be applied in daily life due to a large amount of calculation and poor detection accuracy.To solve the above problems,this paper proposes a dense spatial-temporal graph convolutional network based on lightweight OpenPose.Lightweight OpenPose uses MobileNet as a feature extraction network,and the prediction layer uses bottleneck-asymmetric structure,thus reducing the amount of the network.The bottleneck-asymmetrical structure compresses the number of input channels of feature maps by 1×1 convolution and replaces the 7×7 convolution structure with the asymmetric structure of 1×7 convolution,7×1 convolution,and 7×7 convolution in parallel.The spatial-temporal graph convolutional network divides the multi-layer convolution into dense blocks,and the convolutional layers in each dense block are connected,thus improving the feature transitivity,enhancing the network’s ability to extract features,thus improving the detection accuracy.Two representative datasets,Multiple Cameras Fall dataset(MCF),and Nanyang Technological University Red Green Blue+Depth Action Recognition dataset(NTU RGB+D),are selected for our experiments,among which NTU RGB+D has two evaluation benchmarks.The results show that the proposed model is superior to the current fall detection models.The accuracy of this network on the MCF dataset is 96.3%,and the accuracies on the two evaluation benchmarks of the NTU RGB+D dataset are 85.6%and 93.5%,respectively.

Novel Path Counting-Based Method for Fractal Dimension Estimation of the Ultra-Dense Networks

Next-generation networks,including the Internet of Things(IoT),fifth-generation cellular systems(5G),and sixth-generation cellular systems(6G),suf-fer from the dramatic increase of the number of deployed devices.This puts high constraints and challenges on the design of such networks.Structural changing of the network is one of such challenges that affect the network performance,includ-ing the required quality of service(QoS).The fractal dimension(FD)is consid-ered one of the main indicators used to represent the structure of the communication network.To this end,this work analyzes the FD of the network and its use for telecommunication networks investigation and planning.The clus-ter growing method for assessing the FD is introduced and analyzed.The article proposes a novel method for estimating the FD of a communication network,based on assessing the network’s connectivity,by searching for the shortest routes.Unlike the cluster growing method,the proposed method does not require multiple iterations,which reduces the number of calculations,and increases the stability of the results obtained.Thus,the proposed method requires less compu-tational cost than the cluster growing method and achieves higher stability.The method is quite simple to implement and can be used in the tasks of research and planning of modern and promising communication networks.The developed method is evaluated for two different network structures and compared with the cluster growing method.Results validate the developed method.

Speech Enhancement via Mask-Mapping Based Residual Dense Network

Masking-based and spectrum mapping-based methods are the two main algorithms of speech enhancement with deep neural network(DNN).But the mapping-based methods only utilizes the phase of noisy speech,which limits the upper bound of speech enhancement performance.Maskingbased methods need to accurately estimate the masking which is still the key problem.Combining the advantages of above two types of methods,this paper proposes the speech enhancement algorithm MM-RDN(maskingmapping residual dense network)based on masking-mapping(MM)and residual dense network(RDN).Using the logarithmic power spectrogram(LPS)of consecutive frames,MM estimates the ideal ratio masking(IRM)matrix of consecutive frames.RDN can make full use of feature maps of all layers.Meanwhile,using the global residual learning to combine the shallow features and deep features,RDN obtains the global dense features from the LPS,thereby improves estimated accuracy of the IRM matrix.Simulations show that the proposed method achieves attractive speech enhancement performance in various acoustic environments.Specifically,in the untrained acoustic test with limited priors,e.g.,unmatched signal-to-noise ratio(SNR)and unmatched noise category,MM-RDN can still outperform the existing convolutional recurrent network(CRN)method in themeasures of perceptual evaluation of speech quality(PESQ)and other evaluation indexes.It indicates that the proposed algorithm is more generalized in untrained conditions.

Correg-Yolov3:a Method for Dense Buildings Detection in High-resolution Remote Sensing Images

The exploration of building detection plays an important role in urban planning,smart city and military.Aiming at the problem of high overlapping ratio of detection frames for dense building detection in high resolution remote sensing images,we present an effective YOLOv3 framework,corner regression-based YOLOv3(Correg-YOLOv3),to localize dense building accurately.This improved YOLOv3 algorithm establishes a vertex regression mechanism and an additional loss item about building vertex offsets relative to the center point of bounding box.By extending output dimensions,the trained model is able to output the rectangular bounding boxes and the building vertices meanwhile.Finally,we evaluate the performance of the Correg-YOLOv3 on our self-produced data set and provide a comparative analysis qualitatively and quantitatively.The experimental results achieve high performance in precision(96.45%),recall rate(95.75%),F1 score(96.10%)and average precision(98.05%),which were 2.73%,5.4%,4.1%and 4.73%higher than that of YOLOv3.Therefore,our proposed algorithm effectively tackles the problem of dense building detection in high resolution images.

3D S-wave velocity structure of the Ningdu basin in Jiangxi province inferred from ambient noise tomography with dense array

The Ningdu basin,located in southern Jiangxi province of southwest China,is one of the Mesozoic basin groups which has exploration prospects for geothermal energy.A study on the detailed velocity structure of the Ningdu basin can provide important information for geothermal resource exploration.In this study,we deployed a dense seismic array in the Ningdu basin to investigate the 3D velocity structure and discuss implications for geothermal exploration and geological evolution.Based on the dense seismic array including 35 short-period(5 s-100 Hz)seismometers with an average interstation distance of~5 km,Rayleigh surface wave dispersion curves were extracted from the continuous ambient noise data for surface wave tomographic inversion.Group velocity tomography was conducted and the 3D S-wave velocity structure was inverted by the neighborhood algorithm.The results revealed obvious low-velocity anomalies in the center of the basin,consistent with the low-velocity Cretaceous sedimentary rocks.The basement and basin-controlling fault can also be depicted by the S-wave velocity anomalies.The obvious seismic interface is about 2 km depth in the basin center and decreases to 700 m depth near the basin boundary,suggesting spatial thickness variations of the Cretaceous sediment.The fault features of the S-wave velocity profile coincide with the geological cognition of the western boundary basincontrolling fault,which may provide possible upwelling channels for geothermal fluid.This study suggests that seismic tomography with a dense array is an effective method and can play an important role in the detailed investigations of sedimentary basins.

Possible Nature of an Electron Dense Substance in the Thylakoid Lumen of Chloroplasts

The fixation of leaves of Tanacetum vulgare L. in glutaraldehyde makes it possible to isolate chloroplasts without sacrificing an electron dense substance in the thylakoid lumen. The extraction of lipids from the chloroplasts isolated from the leaves preliminarily fixed in glutaraldehyde has demonstrated that glycerolipids (galactolipids and phospholipids) are not manifested in TLC, whereas isoprenoid lipids (chlorophyll, carotenoids) are manifested. Presumably, isoprenoid lipids are not fixed with glutaraldehyde and are extracted from the thylakoid membrane. The ultrastructural control demonstrates that the electron dense substance from the thylakoid lumen is also extracted. It is possible that this substance is of isoprenoid nature.

An Efficient Quantum Key Distribution Protocol with Dense Coding on Single Photons

Combined with the dense coding mechanism and the bias-BB84 protocol,an efficient quantum key distribution protocol with dense coding on single photons(QDKD-SP)is proposed.Compared with the BB84 or bias-BB84 protocols based on single photons,our QDKD-SP protocol has a higher capacity without increasing the difficulty of its experiment implementation as each correlated photon can carry two bits of useful information.Compared with the quantum dense key distribution(QDKD)protocol based on entangled states,our protocol is more feasible as the preparation and the measurement of a single-photon quantum state is not difficult with current technology.In addition,our QDKD-SP protocol is theoretically proved to be secure against the intercept-resend attack.

Application Research of Dense Sintered Alumina in Iron Runner Castables

Three different kinds of corundum aggregates-tabular sintered alumina, dense sintered alumina, and fused dense corundum-were introduced into the silica fume .free or silica fume containing Al2O3 -SiC - C iron runner castables to investigate their influences on the flow ability, linear change on heating, bulk density, apparent porosity, cold strength, hot modulus of rupture, therm, al shock resistance, slag resistance, oxidation resistance as well as wear resistance of Al2O3 - SiC - C iron runner castables. The results show that （ 1 ） compared with the specimens with fused dense corundum, the specimens with dense sintered alumina have equivalent installation property, slag resistance and oxidation resistance, equivalent or even higher cold modulus of rupture, cold crushing strength and hot modulus of rupture, exhibiting better thermal shock resistance and cold wear resistance ; （2） adopting bimodal alumina micropowder LISAL22RABL as well as water reducers ZX2 and ZD2 can well reduce the water requirement of silica fume free castables, solving the problem of deteriorated flow ability resulted from the lack of silica fume; since the lack of silica fume avoids the formation of low melting point liquid, the hot modulus of rupture and the thermal shock resistance of the silica fume free castables are both better than those of the silica fume containing castables ; （3） the density of the castable specimens with dense sintered alumina is 4% -6% lower than that of the castable specimens with Jhsed dense corundum so the refractories consumption of one iron runner reduces by 5% by using the tastable with dense sintered alumina, which obviously reduces the cost of refractories.

基于ResUNet和Dense CRF模型的地震裂缝识别方法

针对人工解释地震资料耗时长、效率低、受主观因素影响较大的不足,提出了一种基于ResUNet和全连接条件随机场(dense conditional random field, Dense CRF)模型的裂缝识别方法。该方法首先使用ResUNet模型提取地震振幅数据体中裂缝的不同分辨率的特征,实现地震裂缝识别;然后利用Dense CRF模型进一步优化识别结果,从而实现地震裂缝的精准识别。将该方法与传统UNet、ResUNet模型在合成地震振幅数据体和F3工区地震数据体进行了实验比较,结果表明运用所提方法识别的裂缝更准确、裂缝尺寸更细、连续性更好。

Determination of the local magnitudes of small earthquakes using a dense seismic array in the Changning-Zhaotong Shale Gas Field,Southern Sichuan Basin

With the development of unconventional shale gas in the southern Sichuan Basin,seismicity in the region has increased significantly in recent years.Though the existing sparse regional seismic stations can capture most earthquakes with ML≥2.5,a great number of smaller earthquakes are often omitted due to limited detection capacity.With the advent of portable seismic nodes,many dense arrays for monitoring seismicity in the unconventional oil and gas fields have been deployed,and the magnitudes of those earthquakes are key to understand the local fault reactivation and seismic potentials.However,the current national standard for determining the local magnitudes was not specifically designed for monitoring stations in close proximity,utilizing a calibration function with a minimal resolution of 5 km in the epicentral distance.That is,the current national standard tends to overestimate the local magnitudes for stations within short epicentral distances,and can result in discrepancies for dense arrays.In this study,we propose a new local magnitude formula which corrects the overestimated magnitudes for shorter distances,yielding accurate event magnitudes for small earthquakes in the Changning-Zhaotong shale gas field in the southern Sichuan Basin,monitored by dense seismic arrays in close proximity.The formula is used to determine the local magnitudes of 7,500 events monitored by a two-phased dense array with several hundred 5 Hz 3 C nodes deployed from the end of February 2019 to early May 2019 in the Changning-Zhaotong shale gas field.The magnitude of completeness(MC)using the dense array is-0.1,compared to MC 1.1 by the sparser Chinese Seismic Network(CSN).In addition,using a machine learning detection and picking procedure,we successfully identify and process some 14,000 earthquakes from the continuous waveforms,a ten-fold increase over the catalog recorded by CSN for the same period,and the MC is further reduced to-0.3 from-0.1 compared to the catalog obtained via manual processing using the same dense array.The proposed local magnitude formula can be adopted for calculating accurate local magnitudes of future earthquakes using dense arrays in the shale gas fields of the Sichuan Basin.This will help to better characterize the local seismic risks and potentials.

Effect of austempering parameters on microstructure and mechanical properties of horizontal continuous casting ductile iron dense bars

In the present research, the orthogonal experiment was carried out to investigate the influence of different austempering process parameters(i.e. austenitizing temperature and time, and austempering temperature and time) on microstructure and mechanical properties of LZQT500-7 ductile iron dense bars with 172 mm in diameter which were produced by horizontal continuous casting(HCC). The results show that the major factors influencing the hardness of austempered ductile iron(ADI) are austenitizing temperature and austempering temperature. The fraction of retained austenite increases as the austenitizing and austempering temperatures increase. When austenitizing temperature is low, acicular ferrite and retained austenite can be effi ciently obtained by appropriately extending the austenitizing time. The proper austmepering time could ensure enough stability of retained austenite and prevent high carbon austenite decomposition. The optimal mechanical properties of ADI can be achieved with the following process parameters: austenitizing temperature and time are 866 °C and 135 min, and austempering temperature and time are 279 °C and 135 min, respectively. The microstructure of ADI under the optimal austempering process consists of fi ne acicular ferrite and a small amount of retained austenite, and the hardness, tensile strength, yield strength, elongation and impact toughness of the bars are HBW 476, 1670 MPa, 1428 MPa, 2.93% and 25.7 J, respectively.

Reducing environmental risk of nitrogen by popularizing mechanically dense transplanting for rice production in China

The high nitrogen(N)application rates typically used in Chinese cropping systems have led to diminishing returns for yields and have also imposed substantial environmental costs.Here,we estimate that the annual N loss from rice production in China reached approximately 2.6×109 kg from 2011 to 2015,and we demonstrate that adoption of the mechanically dense transplanting technique by producers is an effective method to reduce N loss from rice cropping systems without suffering a yield penalty.

MRI Brain Tumor Segmentation Using 3D U-Net with Dense Encoder Blocks and Residual Decoder Blocks

The main task of magnetic resonance imaging (MRI) automatic brain tumor segmentation is to automaticallysegment the brain tumor edema, peritumoral edema, endoscopic core, enhancing tumor core and nonenhancingtumor core from 3D MR images. Because the location, size, shape and intensity of brain tumors vary greatly, itis very difficult to segment these brain tumor regions automatically. In this paper, by combining the advantagesof DenseNet and ResNet, we proposed a new 3D U-Net with dense encoder blocks and residual decoder blocks.We used dense blocks in the encoder part and residual blocks in the decoder part. The number of output featuremaps increases with the network layers in contracting path of encoder, which is consistent with the characteristicsof dense blocks. Using dense blocks can decrease the number of network parameters, deepen network layers,strengthen feature propagation, alleviate vanishing-gradient and enlarge receptive fields. The residual blockswere used in the decoder to replace the convolution neural block of original U-Net, which made the networkperformance better. Our proposed approach was trained and validated on the BraTS2019 training and validationdata set. We obtained dice scores of 0.901, 0.815 and 0.766 for whole tumor, tumor core and enhancing tumorcore respectively on the BraTS2019 validation data set. Our method has the better performance than the original3D U-Net. The results of our experiment demonstrate that compared with some state-of-the-art methods, ourapproach is a competitive automatic brain tumor segmentation method.