基于改进时空图卷积神经网络的钻杆计数方法

针对现有钻杆计数方法存在劳动重复、计数误差较大、未考虑动作的时序信息等问题,提出了一种基于改进时空图卷积神经网络（MST-GCN）模型的钻杆计数方法。首先,通过矿用监控摄像头获取井下打钻视频数据,采用Alphapose算法在图像序列中提取人体的关键点信息,得到单帧图像上的人体骨架和连续图像序列上的骨架序列数据,进而构建表征人体动作的骨架序列;然后,在时空图卷积神经网络（ST-GCN）模型的基础上设计了MST-GCN模型,采用远空间分区策略关注骨架上距离较远的关键点运动信息,通过注意力机制网络SENet融合原空间特征与远空间特征,从而有效识别骨架序列上的动作类别;最后,在打钻视频上利用支持向量机辨识打钻姿势来决定是否保存骨架序列,若骨架序列长度保存到150帧则使用MST-GCN模型识别动作类别,并根据实际打钻时间设置相邻动作的识别间隔,从而记录动作数量,实现钻杆计数。实验结果表明：在自建的数据集上,MST-GCN模型的识别准确率为91.1%,比ST-GCN、Alphapose-LSTM和NST-GCN动作识别模型的准确率分别提升了6.2%,19.0%和4.8%,模型的损失值收敛在0.2以...     

基于关键点表示的语音驱动说话人脸视频生成

针对现有语音生成说话人脸视频方法忽略说话人头部运动的问题,提出基于关键点表示的语音驱动说话人脸视频生成方法.分别利用人脸的面部轮廓关键点和唇部关键点表示说话人的头部运动信息和唇部运动信息,通过并行多分支网络将输入语音转换到人脸关键点,通过连续的唇部关键点和头部关键点序列及模板图像最终生成面部人脸视频.定量和定性实验表明,文中方法能合成清晰、自然、带有头部动作的说话人脸视频,性能指标较优.     

基于单视频图像序列的人体三维姿态重建

提出了至少存在一个深度值已知点的约束条件下,基于单视频图像序列重建人体三维姿态的方法.利用已知间距的平面点阵来标定获得摄像机参数,在透视投影模型下,根据单视频图像序列中人体关节点的二维数据,重建其三维信息.并将人体运动序列按照运动突变点划分为若干子序列,有效消除了二义性的干扰,较为精确的实现了人体三维姿态的重建.给出了该方法的实验过程及计算结果,验证了该算法的可行性和精度.     

基于轮廓关键点和LSTM的摔倒检测方法

针对传统人体摔倒检测方法误检率高、无法有效结合时序特征等问题,提出一种新的人体轮廓关键点提取方法,并将该方法和LSTM网络相结合构建一种新的摔倒检测模型。该模型对视频中的人体进行轮廓检测,选取轮廓关键点坐标和质心坐标作为人体特征;使用LSTM对人体特征序列进行时序特征提取;用全连接层实现分类。在公开数据集上进行实验,结果表明该模型具有较高的准确率和良好的泛化性。     

低光照场景下基于序列增强的移动机器人人体检测与姿态识别

灾难救援、地下空间开发利用等场景均存在低光照、甚至完全黑暗的问题,导致机器人目标搜索与识别困难。为此,本文面向低光照场景提出基于红外深度相机图像序列的人体检测和姿态识别方法。首先,利用基于YOLO v4的AlphaPose算法检测人体框和关键点。然后,提出基于特征点匹配的漏检人体框恢复算法,降低人体漏检率,同时使用D-S（Dempster-Shafer）证据理论融合人体框和关键点的检测结果,从而降低人体误检率。最后,设计一种基于图像序列信息的人体姿态分层识别方法,在不同的识别层提取不同的人体躯干特征,利用连续多帧躯干向量特征组成的特征序列对人体姿态进行精准的识别并进行实验验证。实验结果表明本文算法能够在低光照条件下实现准确的人体检测与姿态识别,姿态识别准确率高达95.36%。     

基于人体骨骼关键点的吸烟行为检测算法

针对公共场所的监控视频中烟头目标较小并且吸烟产生的烟雾易发散,仅依靠目标检测算法检测烟头或者烟雾来判定吸烟行为存在较大难度的问题,考虑到利用骨骼关键点来进行姿态估计的算法越来越成熟,提出一种利用人体骨骼关键点和吸烟行为之间的关系来进行吸烟行为检测的算法。该算法首先利用AlphaPose和RetinaFace分别检测出人体骨骼关键点和脸部关键点信息,根据手腕到两嘴角中点和手腕到同侧眼睛的距离之比,提出一种计算人体的吸烟动作比例（SAR）是否属于吸烟动作黄金比例（GRSA）的方法以区分吸烟与非吸烟行为;再利用YOLOv4检测视频中是否存在烟头;最后结合GRSA判定和YOLOv4的结果来确定视频中存在吸烟行为的可能性高低,作出是否有吸烟行为的判定。经过笔者录制的数据集测试,结果表明所提算法可以准确检测到吸烟行为,准确率达到92%。     

深度学习方法在二维人体姿态估计的研究进展

人体姿态估计的任务是对图像或视频中的人体关键点进行定位和检测,其一直是计算机视觉领域的热点研究方向之一,也是计算机理解人类行为动作的关键一步。近年来,图像和视频中的二维人体姿态关键点预测在许多领域有着广泛的应用,二维人体姿态估计利用深度学习强大的图像特征提取能力,提升了其鲁棒性、准确性并缩短了处理时间,而且表现效果远超传统方法。根据二维人体姿态研究对象数量的不同,可将其分为单人以及多人姿态估计方法。针对单人姿态估计,根据提取到的关键点表示的不同,可采用基于直接预测人体坐标点的坐标回归方法,以及预测人体关键点高斯分布的基于热图的检测方法;针对多人姿态估计,可采用的方法分为解决多人到单人过程的自顶向下方法,以及直接处理多人关键点的自底向上方法。根据现有的人体姿态估计方法对其进行总结,说明网络结构的内部机制及执行过程,并对常用的数据集、评价指标进行分析,最后阐述当前面临的问题及未来发展趋势。     

基于骨骼关键点的散打动作识别与评价方法

为提高散打运动辅助打分的公平公正性,提出一种基于骨骼关键点的散打动作识别与评价方法。首先采集十类散打动作构建实验数据集,获取视频关键帧,对图像进行去噪处理。其次通过基于YOLOv5s-CBAM目标检测的HRNet-DSC-CBAM人体姿态估计方法提取人体骨骼关键点坐标。最后利用ST-GCN动作识别方法进行动作识别,并利用DTW动态时间规整算法完成动作评价。实验结果表明,该方法在自制数据集的10类散打动作中表现出良好的识别效果,可以实现辅助打分功能。     

一种有效的实时人群计数方法

对固定镜头下视频序列中运动人体的检测和跟踪方法进行研究,利用灰度图像差分双向投影信息检测人体目标,提出一种基于统计运动区域几何特征固定比例的分割算法,使用最近邻匹配方法对人体进行跟踪。完整地实现了一个有效的实时人群计数系统。大量室内和室外场景实验结果表明,该算法具有很好的实时性(每秒处理25帧~30帧且可并行处理4路视频)、对光照变化的鲁棒性以及对稀疏人群检测精度高等特点。     

基于骨骼关键点的异常行为识别及异构平台部署

目前对于异常行为检测算法较多,但是存在检测精度低、对环境要求高、部署困难等缺点。针对以上存在的问题,该文提出了一种基于骨骼关键点的异常行为检测方法。首先对视频图像预处理,然后通过Associative Embedding算法进行人体关键点的提取。为准确描述人体运动,提出用运动特征矩阵进行人体运动描述,引入SVM分类器利用运动特征矩阵进行行为识别。在HMDB51数据中选取的12类异常行为达到平均91.2%准确率,最后模型在CPU+FPGA异构平台进行加速,达到32 FPS的处理速度。     

Frequency-domain identification of continuous-time ARMA models from sampled data

The subject of this paper is the direct identification of continuous-time autoregressive moving average (CARMA) models. The topic is viewed from the frequency domain perspective which then turns the reconstruction of the continuous-time power spectral density (CT-PSD) into a key issue. The first part of the paper therefore concerns the approximate estimation of the CT-PSD from uniformly sampled data under the assumption that the model has a certain relative degree. The approach has its point of origin in the frequency domain Whittle likelihood estimator. The discrete- or continuous-time spectral densities are estimated from equidistant samples of the output. For low sampling rates the discrete-time spectral density is modeled directly by its continuous-time spectral density using the Poisson summation formula. In the case of rapid sampling the continuous-time spectral density is estimated directly by modifying its discrete-time counterpart.     

基于用电大数据的用电异常状态辨识方法

用电异常状态的辨识是用电环节的重点和难点。本文基于计量自动化系统智能电能表所采集的用电大数据,对用电异常状态辨识方法进行研究。首先,基于用电海量数据及高维随机矩阵理论,研究分析了大维随机矩阵的协方差矩阵特征谱分布;然后,根据矩阵的统计特性提出基于用电大数据矩阵的用电异常状态辨识方法;最后,以贵州实际用电数据为例进行了仿真研究。仿真结果表明该文方法不仅能满足电网对可视性、时效性、可靠性、安全性的迫切要求,而且为数据驱动用电环节智能化、可视化监控提供了新思路。     

Energy efficiency and area spectral efficiency tradeoff for coexisting wireless body sensor networks共存无线体域网的能效和区域频效折中

The coexistence of wireless body sensor networks (WBSNs) is a very challenging problem, due to strong interference, which seriously affects energy consumption and spectral reuse. The energy efficiency and spectral efficiency are two key performance evaluation metrics for wireless communication networks. In this paper, the fundamental tradeoff between energy efficiency and area spectral efficiency of WBSNs is first investigated under the Poisson point process (PPP) model and Matern hard-core point process (HCPP) model using stochastic geometry. The circuit power consumption is taken into consideration in energy efficiency calculation. The tradeoff judgement coefficient is developed and is shown to serve as a promising complementary measure. In addition, this paper proposes a new nearest neighbour distance power control strategy to improve energy efficiency. We show that there exists an optimal transmit power highly dependant on the density of WBSNs and the nearest neighbour distance. Some important properties are also addressed in the analysis of coexisting WBSNs based on the IEEE 802.15.4 standard, such as the impact of intensity nodes distribution, optimal guard zone, and outage probability. Simulation results show that the proposed power control design can reduce the outage probability and enhance energy efficiency. Energy efficiency and area spectral efficiency of the HCPP model are better than those of the PPP model. In addition, the optimal density of WBSNs coexistence is obtained.     

Analysing infrequently sampled animal tracking data by incorporating generalized movement trajectories with kernel density estimation

When analysing the movements of an animal, a common task is to generate a continuous probability density surface that characterises the spatial distribution of its locations, termed a home range. Traditional kernel density estimation (KDE), the Brownian Bridges kernel method, and time-geographic density estimation are all commonly used for this purpose, although their applicability in some practical situations is limited. Other studies have argued that KDE is inappropriate analysing moving objects, while the latter two methods are only suitable for tracking data collected at frequent enough intervals such that an object’s movement pattern can be adequately represented using a space–time path created by connecting consecutive points. This research formulates and evaluates KDE using generalised movement trajectories approximated by Delaunay triangulation (KDE-DT) as a method for analysing infrequently sampled animal tracking data. In this approach, a DT is constructed from a point pattern of tracking data in order to approximate the network of movement trajectories for an animal. This network represents the generalised movement patterns of an animal rather than its specific, individual trajectories between locations. Then, kernel density estimates are calculated with distances measured using that network. First, this paper describes the method and then applies it to generate a probability density surface for a Florida panther from radio-tracking data collected three times per week. Second, the performance of the technique is evaluated in the context of delineating wildlife home ranges and core areas from simulated animal locational data. The results of the simulations suggest that KDE-DT produces more accurate home range estimates than traditional KDE, which was evaluated with the same data in a previous study. In addition to animal home range analysis, the technique may be useful for characterising a variety of spatial point patterns generated by objects that move through continuous space, such as pedestrians or ships.     

Study on Power Tower Extraction and Fast Positioning from Airborne LiDAR Data

Airborne LiDAR has become an important technique for transmission line digitalization,reconstruction and safety inspection.Moreover,accurately and efficiently extracting the position of each tower from massive point clouds is basic and important task for the applications in power industry.In this study,a method was proposed to efficiently extract the point clouds and fast determine the position of power towers using airborne LiDAR data.Firstly,the point clouds of power towers were automatically separated from raw data based on the spatial distribution characteristics of airborne LiDAR data.Secondly,each power tower was efficiently detected using a region\|growing algorithm.Finally,a least square linear fitting method was used to determine the accurate position of each power tower.The new proposed method was applied to several LiDAR data sets in areas with high voltage transmission lines.Results indicated that the integrity of the power towers’ points is up to 91.1%,and the accuracy of center positions is high enough with the medium error of 13.5 cm.Additionally,our study also concluded that the proposed method is robust and applicable even the point density is relatively low.     

基于全息时标量测数据挖掘的配电网设备健康状态诊断分析

大数据、数据挖掘等新技术的出现和进步,为建设智能配电网提供了新的技术手段。为实现对配电网运行状态的真实还原、精细分析和精准预测,详实有效的配电网设备运行数据记录和支撑是关键,研究了配电网全息时标量测数据的变化即存储技术,基于全息时标量测数据研究了配电网设备健康状态诊断的方法,对配电网历史数据以及模型信息等进行了深入的数据挖掘,通过聚类分析、线性回归算法、熵权法等建立了设备状态诊断模型和评价体系,实现了对设备故障评估和预警分析等,为及时发现配电网的薄弱环节,保障配电网设备的安全稳定运行提供了有效手段。系统已在地市供电公司的配电网诊断方法研究与实现项目中得到实际应用,很好地满足了地市供电公司的配电网精益化管理需求。     

稳定分布噪声下非线性PAR系统辨识新方法

描述了稳定分布的谱表示,提出了共变谱密度的概念,得到一种基于自共变序列与共变谱的稳定分布白噪声与有色噪声的概念及其判断标准,对传统意义上的白噪声进行了广义化,依据多项式自回归（PAR）系统模型,对基于稳定白噪声输入的系统输出非线性稳定有色噪声建立其非线性PAR模型,提出基于最小P范数的EIRLP算法对非线性PAR系统进行辨识。模拟和分析表明,这种算法是一种在高斯和分数低阶 稳定分布噪声条件下具有良好韧性的非线性系统辨识方法,是对传统的二阶统计量基础上的系统辨识方法的改造与推广。     

统一载波测控系统下行信道噪声功率谱密度标定依据分析与探讨

统一载波测控系统集跟踪、测距、测速、遥测、遥控、通信、数传等功能于一体;噪声功率谱密度是测控设备下行信道的基本参数之一,反映了信道接收弱信号的能力,调节综合基带输入端的噪声功率谱密度至合理范围是完成航天测控任务的前提;介绍了统一载波测控系统下行信道的基本组成及设计要求;分析了噪声功率谱密度在下行信道中的传递特性,提出了在不同信号电平、不同测控体制下噪声功率谱密度标定的基本依据和原则;结合某型号S频段统一测控系统的工程实际,系统分析了噪声功率谱密度的原理和具体标定方法,该方法根据不同应用环境和测控体制,充分考虑各级增益的分配标准,准确计算各级信号噪声的电平,合理进行噪声功率谱密度标定,以确保系统工作的可靠性和稳定性,为设备性能指标测试和航天测控任务参数设置提供参考和依据。     

公理化模糊共享近邻自适应谱聚类算法

针对传统的谱聚类算法通常利用高斯核函数作为相似性度量,且单纯以距离决定相似性不能充分表现原始数据中固有的模糊性、不确定性和复杂性,导致聚类性能降低的问题。提出了一种公理化模糊共享近邻自适应谱聚类算法,首先结合公理化模糊集理论提出了一种模糊相似性度量方法,利用识别特征来衡量更合适的数据成对相似性,然后采用共享近邻的方法发现密集区域样本点分布的结构和密度信息,并且根据每个点所处领域的稠密程度自动调节参数σ,从而生成更强大的亲和矩阵,进一步提高聚类准确率。实验表明,相较于距离谱聚类、自适应谱聚类、模糊聚类方法和地标点谱聚类,所提算法有着更好的聚类性能。