MPEG-4视频中运动背景下的目标检测算法

针对由运动摄像机捕获的MPEG-4视频流中的运动目标检测问题,提出了一种直接利用压缩视频码流进行全局运动估计的新算法.算法从全局运动估计的基础出发,利用背景宏块运动相似性的特点快速建立背景宏块集合并采用常用的四参数全局运动估计模型估计运动参数.最后,计算运动矢量残差,通过对运动矢量残差的筛选检测运动日标.算法利用MPEG-4码流中蕴含的运动信息.不需要对压缩流完全解码,较大地提高了检测效率;进一步改善了检测效果.实验验证了提出的全局运动估计算法的检测效率和检测效果.     

突扩圆管内液-固两相流固体颗粒运动特性的DPM数值模拟

采用考虑颗粒碰撞的欧拉一拉格朗日数值模拟方法(DPM),对水平突扩圆管中液固两相流固体颗粒的碰撞过程进行了数值计算.在模型中,对液相采用欧拉法建立控制方程,对离散颗粒采用拉格朗日方法模拟.采用硬球模型描述颗粒间的碰撞作用.计算结果表明,该模型可以真实地模拟液固两相流中固体颗粒运动的动态变化过程以及颗粒的非均匀分布特征,从单颗粒层次上提供颗粒的运动信息,这有助于深入研究液固两相流中固体颗粒的运动规律.     

伴随均匀流作用的有限水深三阶三色三向表面张力-重力波之完备对称解

着眼于"多色多向"这一典型的海洋表面波传播特征而以"三色三向"加以基本地刻画,并纳入普遍的波-流相互作用机理和丰富多样的表面张力波效应,由此给出了有限水深三阶三色三向波运动系统的一完备对称解.这是对目前业已存在的经典、现代"单色、多色多向波理论"的一次充分包容和集中反映. 关键词： 三色三向波 表面张力-重力波 波-流相互作用 完备对称解     

基于立体视觉与光流融合的运动目标检测

针对摄像机与被检测目标同时运动时的目标检测问题,提出一种立体视觉与光流融合的运动目标检测算法。结合立体视觉技术设计了光流与自运动估计模型,运用车辆的运动信息和场景的深度信息估计因摄像机运动产生的自运动光流;采用多分辨率细化的Horn算法估计场景的混合光流;对混合光流和自运动光流进行差分运算,剔除背景中静态目标的运动干扰。经过一系列形态学滤波处理获得运动目标完整区域,依据光流的连通性对运动目标标号,并确定位置信息。以典型的交通场景为对象进行分析,实验结果表明该算法能有效地检测出动态背景下的运动目标。     

热对流条件下固液密度比对颗粒沉降运动影响的直接数值模拟

本文应用任意拉格朗日-欧拉(ALE)算法对固液两相流流场中考虑热对流的非等温颗粒在竖直通道中的沉降运动进行了数值模拟.在牛顿流体中通过积分黏性应力和压力获得颗粒的受力跟踪颗粒运动,使用有限元方法数值求解流场的N-S方程和能量方程,模型不需经验假设.通过模拟来研究颗粒沉降的运动规律和热对流下固液密度比对固液两相流的影响作用.结果表明随着固液密度比的增加,颗粒经历了稳定沉降、周期性摆动,不规则摆动等过程;热对流使颗粒的摆动幅度和沉降速度发生变化;热对流对颗粒的影响作用随着固液密度比的增加而减小.     

半开式向心涡轮多物理场耦合优化设计

本文引入叶片几何参数化模型、单向流固耦合算法和均匀试验设计法建立了向心涡轮叶片优化平台。通过对某向心涡轮内三维转捩流场、温度场和叶片固体域的求解,完成了该涡轮的流-热-固多物理场耦合优化设计,获得了最优子午面型线、叶片安装角分布和厚度分布,使其效率有所提高.结果表明:向心涡轮的等熵效率由89.2%提高到89.9%,叶片尾缘泄漏流损失明显降低,最大等效应力和最大径向变形量满足材料拉伸屈服强度和叶顶间隙尺寸要求。     

气-固界面作用物理模型的确定性实现方法

针对最新发展的气-固界面作用物理模型,提出一种确定性计算方法.算法的核心是:针对离散速度空间分布函数,采用确定性算法计算出单轮气-固碰撞作用的散射核函数矩阵,并通过考虑吸引势阱作用和迭代累加多次气-固碰撞作用,物理地反映气体分子入射/反射速度分布函数在气-固界面上的变化.与现有基于Monte Carlo随机采样的实现方...     

水平管内栓塞流气力输送的动理学模拟

波状栓塞流是气力输送中的一种典型流型。为了发展更为准确和细致的栓塞流数学模型,本文引入动理学理论建立气-固两相波状栓塞流动的数学模型,对管道内的波状栓塞流动进行了三维数值研究,再现了流动过程中栓塞和堆积层的形成、运动过程,得到了压力降、空气速度、栓塞长度和堆积层厚度等重要的流动参数;栓塞长度、堆积层体厚度随空气表观速度的变化等一些细微流动特征得到了体现,表明动理学理论能够反映出栓塞流这种非连续高浓度气-固流动的物理本质,具有很好的适用性。本文的研究为栓塞流数值模型的发展引入了一种新的数学工具和思路。     

基于浸入边界-多松弛时间格子玻尔兹曼通量求解法的流固耦合算法研究

针对流固耦合问题,发展了基于浸入边界-多松弛时间格子玻尔兹曼通量求解法(immersed boundary method multi-relaxation-time lattice Boltzmann flux solver,IB-MRT-LBFS)的弱耦合算法.依据多尺度Chapman-Enskog展开,建立不可压宏观方程状态变量和通量与格子玻尔兹曼方程中粒子密度分布函数之间的关系;采用强制浸入边界法处理流固界面使固壁表面满足无滑移边界条件,根据修正的速度求解动量方程力源项;结构运动方程采用四阶龙格-库塔法求解.格子模型与浸入边界法的引入使流固耦合计算可以在笛卡尔网格下进行,无需生成贴体网格及运用动网格技术,简化了计算过程.数值模拟了单圆柱横向涡激振动、单圆柱及串列双圆柱双自由度涡激振动问题.结果表明,IB-MRT-LBFS能够准确预测圆柱涡激振动的锁定区间、振动响应、受力情况以及捕捉尾流场结构形态,验证了该算法在求解流固耦合问题的有效性和可行性.     

基于光流的运动小目标检测算法

运动目标的检测是目标识别与跟踪的关键技术之一。光流技术是一种以物体的运动特征来检测目标的方法,它的提出为运动小目标的检测开辟了新的空间。在一个搜索跟踪系统中使用光流技术检测和跟踪空中小目标,目标大概为5～10个像素,而且背景复杂,相机抖动,普通分割算法无法得到小目标。在目标的运动明显异于背景的情况下,通过利用基于光流的目标检测算法来检测出小目标,同时运用高斯金字塔模型,提高算法的运算速度。试验结果表明提出的基于光流的检测算法在背景运动的红外图像中取得了较好的效果。     

A central Rankine–Hugoniot solver for hyperbolic conservation laws

A numerical method in which the Rankine–Hugoniot condition is enforced at the discrete level is developed. The simple format of central discretization in a finite volume method is used together with the jump condition to develop a simple and yet accurate numerical method free of Riemann solvers and complicated flux splittings. The steady discontinuities are captured accurately by this numerical method. The basic idea is to fix the coefficient of numerical dissipation based on the Rankine–Hugoniot (jump) condition. Several numerical examples for scalar and vector hyperbolic conservation laws representing the inviscid Burgers equation, the Euler equations of gas dynamics, shallow water equations and ideal MHD equations in one and two dimensions are presented which demonstrate the efficiency and accuracy of this numerical method in capturing the flow features.     

A comparison between analytic and numerical methods for modelling automotive dissipative silencers with mean flow

Identifying an appropriate method for modelling automotive dissipative silencers normally requires one to choose between analytic and numerical methods. It is common in the literature to justify the choice of an analytic method based on the assumption that equivalent numerical techniques are more computationally expensive. The validity of this assumption is investigated here, and the relative speed and accuracy of two analytic methods are compared to two numerical methods for a uniform dissipative silencer that contains a bulk reacting porous material separated from a mean gas flow by a perforated pipe. The numerical methods are developed here with a view to speeding up transmission loss computation, and are based on a mode matching scheme and a hybrid finite element method. The results presented demonstrate excellent agreement between the analytic and numerical models provided a sufficient number of propagating acoustic modes are retained. However, the numerical mode matching method is shown to be the fastest method, significantly outperforming an equivalent analytic technique. Moreover, the hybrid finite element method is demonstrated to be as fast as the analytic technique. Accordingly, both numerical techniques deliver fast and accurate predictions and are capable of outperforming equivalent analytic methods for automotive dissipative silencers.     

A New Differential Method Applied to the Study of Arbitrary Cross Section Microstructured Optical Fibers

The present work adapts a recent grating theory called “Fast Fourier factorization” to cylindrical coordinates in order to study microstructured optical fibers (MOFs). Compared with the classical differential method, this new differential method takes into account the truncation of Fourier series and the discontinuities of the fields across the diffracting surface with the help of new factorization rules. The main advantage of this method is that the directrix of the diffracting cylindrical surface is arbitrary and permits anisotropic and inhomogeneous media although its numerical application needs longer computation time, compared with other well-known numerical methods. The S-propagation algorithm is used to avoid numerical contaminations. The numerical results are validated and compared with the well-established Multipole method in the case of a MOF with six circular cylinders. Further, a new cross-sectional profile (with sectorial inclusions) that the Multipole method cannot consider is studied.     

An algorithm for LES of premixed compressible flows using the Conditional Moment Closure model

A novel numerical method has been developed to couple a recent high order accurate fully compressible upwind method with the Conditional Moment Closure combustion model. The governing equations, turbulence modelling and numerical methods are presented in full. The new numerical method is validated against direct numerical simulation (DNS) data for a lean premixed methane slot burner. Although the modelling approaches are based on non-premixed flames and hence not expected to be valid for a wide range of premixed flames, the predicted flame is just 10% longer than that in the DNS and excellent agreement of mean mass fractions, conditional mass fractions and temperature is demonstrated. This new numerical method provides a very useful framework for future application of CMC to premixed as well as non-premixed combustion.     

二维二相不规则网格的油藏数值模拟

本文采用不规则网格的差分方法,对平面油水模型进行了数值模拟,并用IMPES方法计算了两个实例。例1的计算结果与精确解吻合良好,比Pedrosa的局部加密网格法使用的网格数少而且计算更简便。例2为矩形网格难以计算的问题,应用本文方法也得出了较好的结果。这表明本文的不规则网格差分方法是可靠而简便的,它可以有效地应用于复杂外边界和复杂地层结构的油藏模拟问题。     

带源参数的热传导反问题的无网格方法

利用无网格有限点法求带有源参数的一维热传导反问题，推导了相应的离散方程.与其他基于网格的方法相比，有限点法采用移动最小二乘法构造形函数，只需要节点信息，不需要划分网格，用配点法离散求解方程，可以直接施加边界条件，不需要在区域内部求积分，减小了计算量.用有限点法求解热传导反问题具有数值实现简单、计算量小、可以任意布置节点等优点.最后通过算例验证了该方法的有效性.     

基于微波工作室的盘荷波导耦合器设计方法

提出了一种新的盘荷波导耦合器数值设计方法; 通过分析Kyhl提出的耦合器调节原理, 将输入阻抗之间的相差关系转换为耦合器输入波导的边界条件关系, 由此 利用Microwave Studio的EigenMode Solver求解本征值问题来实现对耦合器 调谐、调匹配过程的模拟; 并给出一个实例来说明数值计算方法的可行性和可靠性.     

Numerical method satisfying the first two conservation laws for the Korteweg–de Vries equation

In this paper, we develop a finite-volume scheme for the KdV equation which conserves both the momentum and energy. The main ingredient of the method is a numerical device we developed in recent years that enables us to construct numerical method for a PDE that also simulates its related equations. In the method, numerical approximations to both the momentum and energy are conservatively computed. The operator splitting approach is adopted in constructing the method in which the conservation and dispersion parts of the equation are alternatively solved; our numerical device is applied in solving the conservation part of the equation. The feasibility and stability of the method is discussed, which involves an important property of the method, the so-called Jensen condition. The truncation error of the method is analyzed, which shows that the method is second-order accurate. Finally, several numerical examples, including the Zabusky–Kruskal’s example, are presented to show the good stability property of the method for long-time numerical integration.     

Analytical method for computing the electrostatic field distribution in the reflector of the time-of-flight mass spectrometer

The method of analytical calculation of a nonlinear electrostatic mirror of the time-of-flight mass spectrometer and the numerical method of computation of a nonlinear mirror are analyzed. We present the results of simulating of the devices with nonlinear focusing mirrors, which are calculated analytically and by the numerical method, and of the device with a linear reflectron. It is concluded on the basis of the results of simulation that the proposed analytical method for calculating a nonlinear electrostatic mirror of the time-of-flight mass spectrometer is much better than the existing numerical method.     

A new approach with piecewise-constant arguments to approximate and numerical solutions of oscillatory problems

This paper is devoted to the development of a novel approximate and numerical method for the solutions of linear and non-linear oscillatory systems, which are common in engineering dynamics. The original physical information included in the governing equations of motion is mostly transferred into the approximate and numerical solutions. Therefore, the approximate and numerical solutions generated by the present method reflect more accurately the characteristics of the motion of the systems. Furthermore, the solutions derived are continuous everywhere with good accuracy and convergence in comparing with Runge-Kutta method. An approximate solution is developed for a linear oscillatory problem and compared with its corresponding exact solution. A non-linear oscillatory problem is also solved numerically and compared with the solutions of Runge-Kutta method. Both the graphical and numerical comparisons are provided in the paper. The accuracy of the approximate and numerical solutions can be controlled as desired by the number of terms in the Taylor series and the value of a single parameter used in the present work. Formulae for numerical computation in solving various linear and non-linear oscillatory problems by the new approach are provided in the paper.