3D Face Reconstruction from a Single Image Using a Combined PCA-LPP Method

In this paper, we proposed a combined PCA-LPP algorithm toimprove 3D face reconstruction performance. Principal component analysis(PCA) is commonly used to compress images and extract features. Onedisadvantage of PCA is local feature loss. To address this, various studies haveproposed combining a PCA-LPP-based algorithm with a locality preservingprojection (LPP). However, the existing PCA-LPP method is unsuitable for3D face reconstruction because it focuses on data classification and clustering.In the existing PCA-LPP, the adjacency graph, which primarily shows the connectionrelationships between data, is composed of the e-or k-nearest neighbortechniques. By contrast, in this study, complex and detailed parts, such aswrinkles around the eyes and mouth, can be reconstructed by composing thetopology of the 3D face model as an adjacency graph and extracting localfeatures from the connection relationship between the 3D model vertices.Experiments verified the effectiveness of the proposed method. When theproposed method was applied to the 3D face reconstruction evaluation set,a performance improvement of 10% to 20% was observed compared with theexisting PCA-based method.

Time Series Facebook Prophet Model and Python for COVID-19 Outbreak Prediction

COVID-19 comes from a large family of viruses identied in 1965;to date,seven groups have been recorded which have been found to affect humans.In the healthcare industry,there is much evidence that Al or machine learning algorithms can provide effective models that solve problems in order to predict conrmed cases,recovered cases,and deaths.Many researchers and scientists in the eld of machine learning are also involved in solving this dilemma,seeking to understand the patterns and characteristics of virus attacks,so scientists may make the right decisions and take specic actions.Furthermore,many models have been considered to predict the Coronavirus outbreak,such as the retro prediction model,pandemic Kaplan’s model,and the neural forecasting model.Other research has used the time series-dependent face book prophet model for COVID-19 prediction in India’s various countries.Thus,we proposed a prediction and analysis model to predict COVID-19 in Saudi Arabia.The time series dependent face book prophet model is used to t the data and provide future predictions.This study aimed to determine the pandemic prediction of COVID-19 in Saudi Arabia,using the Time Series Analysis to observe and predict the coronavirus pandemic’s spread daily or weekly.We found that the proposed model has a low ability to forecast the recovered cases of the COVID-19 dataset.In contrast,the proposed model of death cases has a high ability to forecast the COVID-19 dataset.Finally,obtaining more data could empower the model for further validation.

Emfacenet:一种轻量级人脸识别的卷积神经网络

随着计算机技术日益发展,计算机视觉逐渐融入人们的生活,深度卷积神经网络在计算机视觉领域得到了广泛的应用.然而计算资源和内存的限制,为卷积神经网络在嵌入式设备的部署带来了巨大的困难.本文提出了一种新的轻量级的人脸识别的卷积神经网络——Emfacenet,通过在CASIA-WebFace数据集上进行卷积神经网络的训练,并在计算机CPU平台以及嵌入式平台上利用LFW数据集对模型的预测效果分别进行测试,Emfacenet在CPU平台下识别速度分别是Resnet50、Mobilenetv3以及Mobilefacenets这3种模型的2.07倍、1.67倍、1.63倍,在嵌入式平台下识别速度分别56.65倍、2.09倍、3.41倍.而且Emfacenet卷积神经网络模型大小仅为138.1KB,保持较高精度的同时运行效率显著提高,可以适用于嵌入式等硬件资源受限领域来实现人脸识别.

Gaussian Mixture Models for Human Face Recognition under Illumination Variations

The appearance of a face is severely altered by illumination conditions that makes automatic face recognition a challenging task. In this paper we propose a Gaussian Mixture Models (GMM)-based human face identification technique built in the Fourier or frequency domain that is robust to illumination changes and does not require “illumination normalization” (removal of illumination effects) prior to application unlike many existing methods. The importance of the Fourier domain phase in human face identification is a well-established fact in signal processing. A maximum a posteriori (or, MAP) estimate based on the posterior likelihood is used to perform identification, achieving misclassification error rates as low as 2% on a database that contains images of 65 individuals under 21 different illumination conditions. Furthermore, a misclassification rate of 3.5% is observed on the Yale database with 10 people and 64 different illumination conditions. Both these sets of results are significantly better than those obtained from traditional PCA and LDA classifiers. Statistical analysis pertaining to model selection is also presented.

Statistical Model-Based Face Pose Estimation

A robust face pose estimation approach is proposed by using face shape statistical model approach and pose parameters are represented by trigonometric functions. The face shape statistical model is firstly built by analyzing the face shapes from different people under varying poses. The shape alignment is vital in the process of building the statistical model. Then, six trigonometric functions are employed to represent the face pose parameters. Lastly, the mapping function is constructed between face image and face pose by linearly relating different parameters. The proposed approach is able to estimate different face poses using a few face training samples. Experimental results are provided to demonstrate its efficiency and accuracy.

Mathematical model of electric hydraulic and powered support control system at a plough mining face

Given the actual working of a fully mechanized plough at a mining face, we have proposed a formula for running constraints between powered supports and a coal plough under assumed geological conditions of the coal face and, on this basis, established an automatic control model of powered supports for the coal plough face. We introduced the working principle of the powered support control system of the plough at the mining face. We established three advanced characteristics of this control system: response speed, reliability and easy maintenance of the system. As well, we briefly introduced, the principal function of primary and subordinate controllers and the realization of the communication system by a Single Bus. Ten controllers were constructed and tested in our laboratorium. The results show that the control model is practical and meets actual conditions. It provides a theoretical basis for designing a computer control system for a powered support system of a plough at a mining face.

A Combined Discrete Event—Agent Based Approach to Modeling Tensile Strength of One-Dimensional Fibrous Materials. Face Validation and Effect of the Basic Model Parameters

A combined method of discrete event and agent based modelling has been applied to the computer modelling and simulation of the tensile strength of one-dimensional fibrous materials (ODFM). This combined method is based on the concept of discrete event simulation as being applied to the modeling of the structure of the fiber flow and on the concept of agent based modelling for modelling and simulation of the fiber interaction within the structure of the fibrous material. Frictional and traction forces arise as the result of this fiber interaction. A model of the ODFM tensile strength, which is based on the slippage effect, is created and studied in this research. Only frictional and traction forces determine the tensile strength in this kind of the model. The article examines the validation problem of the slippage effect based tensile strength model and questions regarding the strength potential estimation through variation in the parameters of the model.

A simplified physically-based breach model for a high concrete-faced rockfill dam:A case study

A simplified physically-based model was developed to simulate the breaching process of the Gouhou concrete-faced rockfill dam （CFRD）, which is the only breach case of a high CFRD in the world. Considering the dam height, a hydraulic method was chosen to simulate the initial scour position on the downstream slope, with the steepening of the downstream slope taken into account; a headcut erosion formula was adopted to simulate the backward erosion as well. The moment equilibrium method was utilized to calculate the ultimate length of a concrete slab under its self-weight and water loads. The calculated results of the Gouhou CFRD breach case show that the proposed model provides reasonable peak breach flow, final breach width, and failure time, with relative errors less than 15% as compared with the measured data. Sensitivity studies show that the outputs of the proposed model are more or less sensitive to different parameters. Three typical parametric models were compared with the proposed model, and the comparison demonstrates that the proposed physically-based breach model performs better and provides more detailed results than the parametric models.

CYCLIC DEFORMATION OF FACE CENTERED CUBIC CRYSTALS AND ITS DISLOCATION INTERACTION MODEL——Ⅱ.DISLOCATION INTERACTION MODEL OF CYCLIC DEFORMATION

A dislocation interaction model has been proposed for cyclic deformation of fcc crystals.Ac- cording to this model,cyclic stress-strain responses and saturation dislocation structures of a crystal are associated with the modes and intensities of dislocation interactions between slip systems active in the crystal; and,hence,may be predicted by the location of its tensile axis in the crystallographic triangle.This model has successfully explained the different behaviours of double-slip crystals and multi-slip behaviours of some crystals with orientations usually con- sidered as single-slip ones.

Seepage simulation of high concrete-faced rockfill dams based on generalized equivalent continuum model

This research focused on the three-dimensional(3 D) seepage field simulation of a high concrete-faced rockfill dam(CFRD) under complex hydraulic conditions. A generalized equivalent continuum model of fractured rock mass was used for equivalent continuous seepage field analysis based on the improved node virtual flow method. Using a high CFRD as an example, the generalized equivalent continuum range was determined, and a finite element model was established based on the terrain and geological conditions, as well as structural face characteristics of the dam area. The equivalent seepage coefficients of different material zones or positions in the dam foundation were calculated with the Snow model or inverse analysis. Then, the 3 D seepage field in the dam area was calculated under the normal water storage conditions, and the corresponding water head distribution, seepage flow, seepage gradient, and seepage characteristics in the dam area were analyzed. The results show that the generalized equivalent continuum model can effectively simulate overall seepage patterns of the CFRD under complex hydraulic conditions and provide a reference for seepage analysis of similar CFRDs.

基于ThetaMEX全局池化的人脸识别神经网络——ShuffaceNet

针对目前大规模网络不适合在手机、平板电脑等资源匮乏的移动设备上使用,以及池化层会导致特征图的稀疏性最终影响神经网络识别精度的问题,提出了一个轻量级人脸识别神经网络ShuffaceNet,设计了一个非线性平滑Log-Mean-Exp函数ThetaMEX,并提出了一种端到端可训练的ThetaMEX全局池化层(TGPL),从而在保证算法精度的前提下,减少网络参数、提高运算速度,进而达到有效地将该网络部署在资源匮乏的移动设备上的目的。ShuffaceNet约有3 600个参数,模型大小仅为3.5 MB。在LFW(Labled Faces in the Wild)、AgeDB-30 (Age Database-30)、CFP (Celebrities in Frontal Profile)人脸数据集上的识别测试的结果表明,ShuffaceNet的精度分别达到了99.32%、93.17%、94.51%。与MobileNetV1、SqueezeNet、Xception相比,所提网络的大小分别缩减了73.1%、82.1%、78.5%,在AgeDB-30数据集上的精度分别提高了5.0%、6.3%、6.7%。可见,基于ThetaMEX全局池化的所提网络能够提高模型精度。

A New Modified Conductivity Model for Prediction of Shear Yield Stress of Electrorheological Fluids Based on Face-center Square Structure

A new modified conductivity model was established to predict the shear yield stress of electrorheological fluids (ERF). By using a cell equivalent method, the present model can deal with the face-center square structure of ERF. Combining the scheme of the classical conductivity model for the single-chain structure, a new formula for the prediction of the shear yield stress of ERF was set up. The influences of the separation distance of the particles, the volume fraction of the particles and the applied electric field on the shear yield stress were investigated.

Research on 3D Modeling and Visualization of Coal Pillars for Surface Protection

In order to safely exploit coal resource, protection coal pillars must be prepared in coal mines. Some correlative parameters of protection coal pillar are calculated by Drop face and Drop line methods. Models of protecting surface objects and coal pillars are established by TIN modeling and object-oriented technique. By using ACCESS2000as the database and the VC++ and OpenGL as the language, the calculation of protective coal pillars is realized and the 3D-visulizaiton system for protected objects on ground surface and for coal pillars is developed. The system can obtain the data of characteristic points on the surface interactively from the digitized mine topography map, constructing 3D model automatically. It can also obtain the interrelated parameters of the coal seam and drill hole data from existing geological surveying database to calculate the location, surface area and the total coal columns. The whole process can be computed quickly and accurately. And the 3D visualization system was applied in a mine, showing that the system solve the problem of complex calculation,not only realized the automatic 3D mapping and visualization of coal pillars for buildings protection , but also greatly improves the working efficiency.

Mode-I-crack compression modeling and numerical simulation for evaluation of in-situ stress around advancing coal workfaces

The relatively high stress probably leads to generation of a fractured or even instable area around a working coalface. Also, the generated weak area often evolves into an easy-infiltrating field of water/gas to greatly increase probability of accident occurrence. To reveal the distribution of high stress around working faces, we put forward the mode-I-crack compression model. In this model, the goaf following a working face is regarded as a mode-I crack in an infinite plate, and the self-gravity of overlaying strata is transformed into an uniform pressure applied normal to the upper edge of the model crack. Solving this problem is based on the Westergaard complex stress function. For comparison, the software RFPA-2D is also employed to simulate the same mining problem, and furthermore extendedly to calculate the stress interference induced by the simultaneous advances of two different working faces. The results show that, the area close to a working face or the goaf tail has the maximum stress, and the stress is distributed directly proportional to the square root of the advance and inversely proportional to the square root of the distance to the working face. The simultaneous advances of two neighboring working faces in different horizontals can lead to extremely high resultant stress in an interference area.

基于特征融合的轻量级新残差人脸识别方法

针对现有轻量级模型在嵌入式设备的人脸识别应用中存在识别精度难以提升的问题,提出一种融合人脸对齐关键特征点信息的轻量级新残差网络模型(Lightweight New Residual Network,LNRN).LNRN利用深度残差网络结构能够解决网络退化且避免干扰因素影响的优势,结合人脸对齐环节产生的关键特征点信息,对深度残差网络结构进行简化和合理设计,实现对关键特征信息和全局信息的提取.为避免特征提取过程中丢失重要特征信息,该模型在新残差网络中加入结合空间和通道的注意力机制进行辅助.在公开的四个标准人脸数据集上的仿真实验表明,该模型识别速度在接近主流轻量级人脸识别方法的同时,平均识别精度比MobiFace提高了0.6%.

基于边塌陷减面的实景三维模型轻量化技术

实景三维模型由于其丰富的场景展现能力,广泛应用在数字孪生和地理测绘等领域。然而,丰富的三维场景和细节所包含的数据量十分庞大,给实景三维模型的存储、传输、渲染与展示带来了极大的挑战。针对上述问题,本文提出了一种面向实景三维模型的轻量化处理方法,首先引入模型预处理技术,去除模型噪音和非流行边;然后通过边塌陷减面的轻量化算法,精简模型拓扑结构;最后加入反转检测算法,防止边塌陷过程中出现面反转的情况。对张家界景区4个真实场景的测试结果表明,该技术可以在保留模型整体外观和场景细节的基础上,压缩至原始模型大小的3.7%~10.9%,使得复杂场景的实景三维模型能够在计算机中更加快速真实地反映或显示世界的地理地貌或城市建筑。

FACE水稻茎蘖动态模型

借助农田开放式空气CO2浓度增高(FACE)技术平台,以武香粳14为供试水稻品种,设置不同施N量处理,研究大气CO2浓度为570μmol·mol-1(比对照高200μmol·mol-1)的FACE处理对水稻茎蘖动态的影响,并建立了相应的模拟模型:Tt=1+Ae1a1-b1t-1+Aea22-b2t+C×1+Ba31-b3t-1+Be2a4-b4t+D.模型以时间为驱动因子,描述了水稻茎蘖数随移栽天数的动态变化过程,对常规及CO2浓度增加条件下水稻茎蘖的变化均有很好的拟合性.通过不同年份试验数据对模型的检验,预测根均方差(RMSE)最大为44·27个·m-2,最小为13·96个·m-2,且相关系数均达到了极显著水平.表明模型的预测程度较高,具有很好的适用性.

海底软弱地层浅埋大直径盾构对接开挖面失稳灾变机制研究

为探明对接距离对海底大直径盾构开挖面失稳机制及规律的影响,依托甬舟铁路金塘海底盾构隧道工程,建立考虑渗流的盾构隧道对接有限元模型,研究对接距离L对开挖面失稳机制以及极限支护压力的影响,并基于极限平衡法建立考虑渗流的盾构近距离对接开挖面极限支护压力理论模型。随着对接距离的增大,开挖面失稳形状和极限支护压力的变化可分为3个阶段。1)快速增长阶段(0<L/D≤0.4)(D为隧道直径):该阶段与单个隧道工况区别最明显,失稳区域由楔形体与仓筒组成;此时受对接隧道的阻碍作用最大,且开挖面附近水头场变化剧烈,渗流力占据主要部分。2)缓慢增长阶段(0.4<L/D≤0.7):失稳区域由对数螺旋体与仓筒组成,此时对接隧道的阻碍作用减小,因此极限支护压力在该阶段受对接距离的影响减小,增长缓慢。3)趋于平缓阶段(0.7<L/D≤1.5):失稳区域由对数螺旋体与倒棱台组成,此阶段对接隧道的阻碍作用可忽略不计,极限支护压力、失稳区域形状与非对接情况相近,可视为单个隧道工况。建立的理论模型得出的结果与数值模拟的结果接近,两者误差在10%以内,验证了该模型的可行性。