Off-Line Signature Recognition Based on Angle Features and Artificial Neural Network Algorithm

Handwritten signature recognition is presented based on an angle feature vector by using the artificial neural network （ANN） in this research. Each signature image will be represented by an angle vector. The feature vector will constitute the input to the ANN. The collection of signature images is divided into two sets. One set will be used for training the ANN in a supervised fashion. The other set which is never seen by the ANN will be used for testing. After training, the ANN will be tested by recognizing the signatures. When a signature is classified correctly, it is considered correct recognition, otherwise it is a failure. The achieved recognition rate of this system is 94%.

Japanese Sign Language Recognition by Combining Joint Skeleton-Based Handcrafted and Pixel-Based Deep Learning Features with Machine Learning Classification

Sign language recognition is vital for enhancing communication accessibility among the Deaf and hard-of-hearing communities.In Japan,approximately 360,000 individualswith hearing and speech disabilities rely on Japanese Sign Language(JSL)for communication.However,existing JSL recognition systems have faced significant performance limitations due to inherent complexities.In response to these challenges,we present a novel JSL recognition system that employs a strategic fusion approach,combining joint skeleton-based handcrafted features and pixel-based deep learning features.Our system incorporates two distinct streams:the first stream extracts crucial handcrafted features,emphasizing the capture of hand and body movements within JSL gestures.Simultaneously,a deep learning-based transfer learning stream captures hierarchical representations of JSL gestures in the second stream.Then,we concatenated the critical information of the first stream and the hierarchy of the second stream features to produce the multiple levels of the fusion features,aiming to create a comprehensive representation of the JSL gestures.After reducing the dimensionality of the feature,a feature selection approach and a kernel-based support vector machine(SVM)were used for the classification.To assess the effectiveness of our approach,we conducted extensive experiments on our Lab JSL dataset and a publicly available Arabic sign language(ArSL)dataset.Our results unequivocally demonstrate that our fusion approach significantly enhances JSL recognition accuracy and robustness compared to individual feature sets or traditional recognition methods.

Effect of Cone Angle on the Hydraulic Characteristics of Globe Control Valve

Globe control valve is widely used in chemical, petroleum and hydraulic industries, and its throttling feature is achieved by the adopting of valve plug. However, very limited information is available in literature regarding the influence of valve plug on the internal and external features in globe control valves. Thus the effect of valve plug is studied by CFD and experiment in this paper. It is obtained from external features that the pressure drop between upstream and downstream pressure-sampling position increases exponentially with flow rate. And for small valve opening, the increment of pressure drop decreases with the increase of cone angle（β）. However, with the increase of valve opening, the effect of cone angle diminishes significantly. It is also found that the cone angle has little effect on flow coefficient（Cv） when the valve opening is larger than 70%. But for the cases less than 70%, Cv curve varies from an arc to a straight line. The variation of valve performance is caused by the change of internal flow. The results of internal flow show that cone angle has negligible effect on flow properties for the cases of valve opening larger than 70%. However, when valve opening is smaller than 70%, the pressure drop of orifice decreases with the increase of β, making the reduction in value and scope of the high speed zone around the conical surface of valve plug, and then results in a decreasing intensity of adjacent downstream vortex. Meanwhile, it is concluded from the results that the increase of cone angle will be beneficial for the anti-cavitation and anti-erosion of globe control valve. This paper focuses on the internal and external features of globe control valve that caused by the variation of cone angle, arriving at some results beneficial for the design and usage of globe control valve.

Feature Patch Illumination Spaces and Karcher Compression for Face Recognition via Grassmannians

Recent work has established that digital images of a human face, when collected with a fixed pose but under a variety of illumination conditions, possess discriminatory information that can be used in classification. In this paper we perform classification on Grassmannians to demonstrate that sufficient discriminatory information persists in feature patch (e.g., nose or eye patch) illumination spaces. We further employ the use of Karcher mean on the Grassmannians to demonstrate that this compressed representation can accelerate computations with relatively minor sacrifice on performance. The combination of these two ideas introduces a novel perspective in performing face recognition.

Analysis on the Writing Characteristic Featuresof The Great Gatsby

The Great Gatsby is considered the masterpiece of F.Scott Fitzgerald.The reason why it is one of the greatest novels all over the world is not only because of its showing the cruel social reality in America during 1920s and the selflessness and merciless of human nature at that time,but also the novel's unique narrative angle that Fitzgerald wrote the novel from the first person perspective.Fitzgerald com posed the figure "Nick Carraway" as the narrator of the story of Gatsby.And the plots of the novel are ingeniously conceived and well-knit.It gets coherence between the beginning and the end,and the plots linked together.So the novel makes a strong impression on readers.

Adaptive Human Tracking Across Non-overlapping Cameras in Depression Angles

To track human across non-overlapping cameras in depression angles for applications such as multi-airplane visual human tracking and urban multi-camera surveillance,an adaptive human tracking method is proposed,focusing on both feature representation and human tracking mechanism.Feature representation describes individual by using both improved local appearance descriptors and statistical geometric parameters.The improved feature descriptors can be extracted quickly and make the human feature more discriminative.Adaptive human tracking mechanism is based on feature representation and it arranges the human image blobs in field of view into matrix.Primary appearance models are created to include the maximum inter-camera appearance information captured from different visual angles.The persons appeared in camera are first filtered by statistical geometric parameters.Then the one among the filtered persons who has the maximum matching scale with the primary models is determined to be the target person.Subsequently,the image blobs of the target person are used to update and generate new primary appearance models for the next camera,thus being robust to visual angle changes.Experimental results prove the excellence of the feature representation and show the good generalization capability of tracking mechanism as well as its robustness to condition variables.

On Electromagnetic Scattering Features of Geological Radar Targets

The electromagnetic scattering principles of geological radar targets and various influential factors were discussed, and the importance of researching into the electromagnetic scattering features of the targets to the actual prospecting task was pointed out.

OLEOPHOBIC AND HYDROPHOBIC FEATURE EXPERIMENTS OF FLUORINATED HIGH DENSITY POLYETHYLENE

The surface performances of directly fluorinated high density polyethylene （HDPE） are studied with Fourier transform infrared （FT-IR） spectra, scanning electron microscopy （SEM） and contact angle （CA） system. The SEM images show that there is a three-layer structure called the reaction, virgin and boundary layer structure. The depth of fluorinated layer is 5.75 ~m with 1 h fluorination time and 7.86 b^m with 2 h. The depths are 5.46 /~m and 5.07 /~m when fluorine density is 2G and 1~/0, respectively. CA indicates that the HDPE surface property becomes more hydrophobic with the increasing water contact angle from 78.5~ to 104.5~. Oleophobic and hydrophobic features of HDPE are identified by comparison of mass change experiments. It is shown that the in- crement rate of fluorinated HDPE is much lower than that of un-fluorinated HDPE filled in neither distilled water nor jet fuel.

基于spike-angle图和Fourier变换的磨粒图像形状特征提取

根据磨粒Spike参数的思想,提出了spike-angle参数和spike-angle图的概念,以及基于spike-angle图的磨粒形状特征提取方法。首先计算磨粒的所有轮廓点在不同步长时的spike-angle参数,得到磨粒的spike-angle图,spike-angle图的行反映了磨粒轮廓的整体特征,spike-angle图的列反映了磨粒轮廓的局部特征;然后对spike-angle图的行和列分别进行傅里叶变换,并对傅里叶变换系数幅值进行归一化处理;最后每一阶归一化的傅里叶变换系数幅值进行直方图统计,以归一化的整体直方图特征和局部直方图特征作为磨粒的形状特征。论文提出的磨粒形状特征提取方法充分利用了磨粒形状的整体信息、局部信息和多尺度信息,并且具有平移、尺度和旋转不变性。磨粒类型识别实验结果表明,所提的方法能够有效地用于磨粒类型识别。

基于转向架航向角的既有铁路平面线形识别方法研究

轨道平面线形识别是解算轨道不平顺的基础。针对外挂式动态轨道检查仪的实际需要,提出一种基于转向架航向角的铁路平面线形识别方法。首先,通过分段试拟合判定准则自动识别各测点所属线元类型和特征点数量;然后,建立基于转向架航向角的平面线形特征点优化模型;最后,结合云模型改进传统遗传算法的搜索策略提出云遗传算法,实现铁路平面线形精确识别。研究表明,平面线路曲线特征点识别偏差在4 m以内,线形识别准确度均达到97%以上,满足外挂式动态轨道检查仪对轨道不平顺解算的精度要求。

基于I-ω 响应关联特征的新能源电力系统暂态功角稳定判别方法

基于响应驱动的暂态功角稳定分析方法具有不依赖模型和参数、计算量小、能够实时量测和实时判别等突出优势。首先,该文针对单机无穷大系统挖掘出发电机电流和角速度蕴含的暂态功角稳定信息,建立了故障后电流与角速度的关联关系,揭示了两电气量存在的暂态稳定性特征;其次,利用信息量测技术得到发电机电流与角速度的关系曲线,通过分析关系曲线特性,提取出系统功角稳定和功角失稳关键特征,并探究了风机接入后对关系曲线特性的影响,提出了一种暂态功角稳定新判据;最后,通过比较故障清除后机端母线电压相位差变化率的大小识别临界机群,将该文方法应用于多机系统暂态稳定性判别。仿真算例结果表明了所提方法能够简单快速地判断出暂态功角稳定态势。

机载LiDAR系统安置角自适应迭代修正算法探讨

安置角参数在机载激光雷达作业中对后期点云数据精度质量有着极大的影响,使其真实地物点位置发生改变,相邻航带点云出现分层,而传统布设地面控制点的标定模式效率低、成本高、要求高、可操作性差,为规避上述诸多弊端,给出一种安置角自适应的标定模型,通过相邻重叠航带内部大量同名特征点的空间位置匹配、安置角参数的迭代运算,利用最小二乘配置方法求解机载激光雷达扫描系统的三个安置角参数,并借助C语言实现算法程序,在测试案例中得到了验证,提升了安置角标定效率,提高了机载激光雷达技术在复杂地形中点云的处理质量、适用性与可操作性。

改进黑翅鸢算法的1D-2D-GAF-PCNN-GRU-MSA弓网电弧检测应用

针对高速列车运行时高速气流场对受电弓碳滑板与接触网之间的接触压力和电弧状态影响。通过计算得出更符合实际状态下的接触压力和电弧状态模型,建立了考虑高速气流场影响的弓网电弧实验模型。提出了改进黑翅鸢算法(IBKA)的1D-2D-GAF-PCNN-GRU-MSA故障检测模型。应用格拉姆角场(GAF)将一维接触电压信号时序图象化转换为二维图像并通过双通道卷积神经网络(PCNN)进行特征识别。另将一维时序信号通过门控循环单元(GRU)捕捉时序信号特征。将一维时序信号特征与二维图像特征进行特征融合,弥补各自局限性。针对模型中的难以确定的学习率、门控循环单元网络层神经元个数等参数,融入改进黑翅鸢算法(IBKA)对参数寻优使模型更加合理。最后,融合多头自注意力机制提高模型准确率。将提出的模型与其他3种模型分别对3组不同实验条件的弓网电弧模型进行检测,验证提出的模型具有较强的鲁棒性和较高的准确性。

基于改进YOLOv7-tiny的大尺寸导光板缺陷检测

针对导光板缺陷种类多、尺寸小、人工检测效率低的问题,提出一种基于改进YOLOv7-tiny的大尺寸导光板缺陷检测方法。首先,通过对导光板图像进行滑窗剪切以解决图像分辨率过大的问题;然后,对小样本缺陷使用多角度数据增强技术丰富数据量以解决样本不均衡的问题;最后,将轻量级卷积注意力模块(CBAM)整合到YOLOv7-tiny主干特征提取部分,使模型在通道和空间上对小目标缺陷的特征提取能力得到增强;选取WIoUv2损失函数计算定位损失,增强网络对困难示例的关注度,提高算法对低质量锚框的检测能力。实验结果表明,所提方法的均值平均精度为85.8%、召回率为81.3%,与原始YOLOv7-tiny相比,分别提高了5.4%和8.1%。

下肢运动状态特征对裤装臀围的影响分析

为达成合体不束缚目标,探究裤装在不同运动状态下的臀围需求,综合分析人体特征参数、下肢动态参数的约束关系,基于单腿、双腿抬起为起点探讨臀围截面形态变化规律,构建了臀围与人体特征参数、下肢动态参数之间的数学模型,并进行趋势分析以及模型修正。研究结果表明:在单侧抬腿状态时,臀围会随着方向角度α的增加呈现先减小后增大的趋势;而在双侧抬腿状态时,其与臀围呈现负相关关系。抬起角度β与臀围呈正相关,在抬起角度β≤60°时,可按人体运动需求角度计算裤装臀围值;在β>60°时,需要使用坐宽参数代替臀围宽度参数。在下肢多种运动状态下选取臀围变化后的最大值作为不束缚人体的裤装臀围最小值。模型解决了下肢运动状态对裤装臀围的影响,可为裤装臀围制版提供理论依据,为服装定制提供理论参考。

基于光场解耦的6D位姿估计方法

光场成像技术能同时捕捉场景中光线的空间和角度信息,被广泛应用到许多计算机视觉任务中。针对基于RGB图像位姿估计方法在严重遮挡和截断、光照变化、物体和背景相似等复杂场景下难以有效预估出位姿的问题,提出一种光场解耦特征融合的两阶段6D位姿估计方法。该方法采用多种特征提取器解耦光场宏像素图像并将其映射到特征空间,并引入注意力机制融合空间角度及EPI信息,为下游位姿估计网络提供有效可靠的关键特征。同时,将反投影应用到关键点预测网络以减少特征传递过程中信息的损耗。在光场位姿估计数据集LF-6Dpose上的实验表明,该方法在平均最近点三维距离(ADD-S)和二维投影(2D Project)两个指标下的结果分别为91.37%和70.12%,在三维距离指标上相比现有先进方法提升12.5%,能够更好地解决复杂场景下的目标6D位姿估计问题。

幼童脑干形态学与增龄发育的数字化特征

背景:既往对脑干研究多集中于成人和胎儿,而对国人幼童脑干的增龄发育特征研究甚少。目的:观测健康幼童脑干发育特征,探究其增龄差异及其与性别间的相关性。方法:2019年1月至2022年4月对内蒙古医科大学附属医院174例2-6岁幼童颅颈部3.0T MRI图像行回顾性研究,测量脑干(含中脑、脑桥、延髓)正中矢状面径线、面积和角度。结果与结论:中脑、脑桥、延髓前后径在2-5岁组,中脑、脑桥、延髓纵径和面积在2-6岁组总体随增龄呈递增趋势,除延髓纵径外均呈正相关(r>0,P<0.05),在2-3岁组、4-5岁组期间处于快速生长发育阶段,这2个年龄组可作为幼童发育重点观测指标;脑桥前后径、纵径、面积和脑干总面积与年龄相关性较强,可作为其脑干发育重点观测指标。

基于改进KCF算法的织物折皱回复检测研究

织物折皱回复性能是评价织物形态稳定性的关键指标。传统折皱回复角测试方法存在检测过程依赖人工操作、难以量化折皱动态演变等问题。为实现对折皱回复全过程的自动化监测,文章提出一种基于改进核相关滤波算法的动态折皱回复检测方法。该方法使用高速摄像机捕捉织物折皱形变回复过程,应用改进的核相关滤波算法检测追踪折皱顶点的运动角度变化。通过引入多特征融合提高检测鲁棒性,利用Canny边缘检测自适应调整目标区域,减小边界效应。在此基础上提取感兴趣区域骨架,计算折皱顶角度随时间变化信息。结果表明,不同织物折皱角度变化规律与织物组织结构高度相关。最后与标准测试结果建立线性模型,验证所提方法的有效性。文章实现了对织物折皱回复全过程的自动化检测与定量评估,提供了一种更为高效准确的折皱回复性能检测新思路,具有广阔应用前景。

小角中子散射物理模型自动化筛选

小角中子散射(SANS)实验数据的分析过程需要科研人员选择样品对应的物理模型进行迭代拟合来表征样品的结构和属性。目前选择物理模型的方法大多是基于人工经验,分析门槛高、准确率较低。基于标准神经网络模型的小角中子散射实验样品物理模型自动化筛选方法面临着图像缺乏局部特征、类内差异大、类间差异小等问题。设计双模态特征融合卷积神经网络(BFF-CNN)模型,先引入物理感知的傅里叶-贝塞尔变换(FBT)来提取散射图像的全局结构信息,再将原始图像与FBT变换图像通过两个子网络分别进行特征提取与特征融合,以提升神经网络整体的特征表达能力。提出受限Softmax(R-Softmax)损失函数,通过在原生Softmax损失函数的基础上添加惩罚项来限制输入样本被分配到非本真类的概率,可在Softmax损失接近0时缓解梯度的消失问题,进而提高收敛速度。在自建的小角中子散射图像数据集上的实验结果表明,BFF-CNN的预测准确率和平均召回率相比于ResNet-18、PMG等模型提升幅度较大,采用R-Softmax与中心损失函数的联合学习策略后的预测准确率和召回率相比只采用Softmax损失函数提升了5.4和10.5个百分点,具有较好的小角中子散射图像分类效果。

改进丰富卷积特征算法的液滴边缘检测模型

液滴图像边缘的高精度提取是测量水接触角较为关键的一环,针对常规边缘提取方法噪声鲁棒性差、边缘提取不完整、精度低的问题,本文提出了一种改进丰富卷积特征(RCF)的液滴边缘检测模型。首先,在深度特征提取阶段引入特征融合模块,使用多个特征让模型更加鲁棒,减少过拟合的风险;其次,设计多感受野模块代替RCF后边的contact层,通过多个感受野来提取更多的语义信息,使边缘细节更加丰富;然后,在模型每一层之前引入高效通道注意力机制,增强模型对图像中重要特征的关注程度;最后,设计并引入MaxBlurPool下采样技术,减少计算量和参数量,提高平移不变性。在自制液滴数据集上的实验结果表明,本文模型的固定轮廓阈值(ODS)提高到0.816、单图像最佳阈值(OIS)提高到0.829、检测准确率高达90.17%,相较原模型提高了1.85个百分点,能够准确检测液滴边缘特征。