A hybrid vertex-centered finite volume/element method for viscous incompressible flows on non-staggered unstructured meshes

This paper proposes a hybrid vertex-centered fi- nite volume/finite element method for solution of the two di- mensional （2D） incompressible Navier-Stokes equations on unstructured grids. An incremental pressure fractional step method is adopted to handle the velocity-pressure coupling. The velocity and the pressure are collocated at the node of the vertex-centered control volume which is formed by join- ing the centroid of cells sharing the common vertex. For the temporal integration of the momentum equations, an im- plicit second-order scheme is utilized to enhance the com- putational stability and eliminate the time step limit due to the diffusion term. The momentum equations are discretized by the vertex-centered finite volume method （FVM） and the pressure Poisson equation is solved by the Galerkin finite el- ement method （FEM）. The momentum interpolation is used to damp out the spurious pressure wiggles. The test case with analytical solutions demonstrates second-order accuracy of the current hybrid scheme in time and space for both veloc- ity and pressure. The classic test cases, the lid-driven cavity flow, the skew cavity flow and the backward-facing step flow, show that numerical results are in good agreement with the published benchmark solutions.

Simulation of a ground-layer adaptive optics system for the Kunlun Dark Universe Survey Telescope

Ground Layer Adaptive Optics （GLAO） is a recently developed technique extensively applied to ground-based telescopes, which mainly compensates for the wavefront errors induced by ground-layer turbulence to get an appropriate point spread function in a wide field of view. The compensation results mainly depend on the turbu-lence distribution. The atmospheric turbulence at Dome A in the Antarctic is mainly distributed below 15 meters, which is an ideal site for applications of GLAO. The GLAO system has been simulated for the Kunlun Dark Universe Survey Telescope, which will be set up at Dome A, and uses a rotating mirror to generate several laser guide stars and a wavefront sensor with a wide field of view to sequentially measure the wavefronts from different laser guide stars. The system is simulated on a computer and parameters of the system are given, which provide detailed information about the design of a practical GLAO system.

掺锶的偏硼酸钡(Sr_xBa_(1-x)B_2O_4)大块单晶的生长和表征

Barium metaborate (BaB 2O 4) exists in two forms the high temperature for m α phase and the low temperature form β phase,with the phase transition poi nt at around 925℃.The low temperature phase (β BBO) is an excellent NLO cryst al for UV region,while the high temperature one (α BBO) is known as a good bir efringent material.α BBO crystal possesses better transparency in the ultravio let region compared with commonly used YVO 4 and CaCO 3 crystals.However,due t o phase transition problem,it is difficult to grow single α BBO crystals by D CZ method,which restricts its wider application.α BBO crystal usually cracks upon cooling resulted from phase transition to β phase.Therefore,it is difficult to grow β BBO single crystals directly from pure BaB 2O 4 mel t;however,to grow single crystals of α BBO is not easy,either. Our recent experiments showed that one could avoid cracking of α BBO cryst als by Sr 2+ doping.This minute amount of Sr 2+ plays the role of structure stabilizer,which inhibits structural reconstruction. In this way,Sr 2+ d op ed BBO single crystals do not subject to phase transition from 925℃ down to roo m temperature. We have mow successfully grown out α BBO single crystals 30mm in diameter by D CZ method from Sr x Ba 1- x BO 4 melt with Sr concentration 0.3 0 .5%.The ratation rate is 10 15r/min,pulling speed is 1 2mm/h.Preliminary tests revealed that its structure and physic chemical properties were almost identic al to those grown from pure melts.These Sr 2+ doped α BBO crystals are no w being widely used in optical isolators.

基于改进L-K光流法的电缆断裂伸长率测量技术

针对现有电缆断裂伸长率测量方法稳定性差、自动化程度低的缺陷提出了一种基于改进L-K光流法的电缆断裂伸长率测量方法。首先,在待测电缆表面的两端,使用一种黑色圆环标记符构建特征点。然后采用边缘提取和圆度计算等方式获取电缆标记符的初始形心位置,并与Harris算法检测出的特征点按欧氏距离最小的原则匹配,从而实现特征点的自动选取。接着在电缆拉伸的过程中,使用改进的L-K光流法追踪特征点并实时计算特征点的加速度和间距。本文提出了一种加速度突变判据,该判据能够准确定位电缆断裂所在帧,从而解决了人工判定电缆断裂瞬间精度差的问题。最后根据电缆形变前后特征点间的距离计算得到电缆的断裂伸长率。实验测得电缆断裂伸长率平均误差为2.78%,算法平均速度为21 frames/s,表明本文算法能够快速准确地测量电缆断裂伸长率。

Design and experimental test of an optical vortex coronagraph

Using an optical vortex coronagraph （OVC） is one of the most promising techniques for di- rectly imaging exoplanets because of its small inner working angle and high throughput. This paper presents the design and laboratory demonstration performance of an OVC based on liquid crystal polymers （LCPs） at 633 nm and 1520 nm. The OVC can deliver good performance in laboratory tests and achieve a contrast of 10-6 at an angular distance of 3A/D, which can be implemented for imaging young giant exoplanets in combination with extreme adaptive optics.

特征点法与光流跟踪融合的视觉里程计算法

针对视觉里程计中特征点法因计算量大而导致实时效果不好的问题,提出了一种特征点法与光流跟踪融合的算法。首先,通过度量运动大小的视差阈值和地图点的跟踪衰减系数作为筛选策略选取关键帧;其次,仅对关键帧中的特征点计算描述符,而对于非关键帧中的特征点则是通过基于图像金字塔的稀疏光流法建立起关键点间的对应关系,从而降低计算量;最后,通过最小化重投影误差得到当前帧的位姿估计。通过实验测试证明,该算法的定位精度与ORBSLAM2相当,但运行速度可提高70%以上,实时性明显优于ORBSLAM2算法。

基于改进光流法的视频监控中人群异常行为检测算法

危害社会公共安全的事件频发,研究视频监控中人群异常行为对于恢复治安秩序和保障公众安全有着重要意义。由于视频监控涉及多种多样的场景,复杂环境影响了人群异常行为的准确检测。因此,为提升视频监控中人群异常行为的检测效果,提出基于改进光流法的视频监控中人群异常行为检测算法。利用改进单高斯模型在视频监控中人群视频帧内提取角点作为特征点;基于改进光流法计算特征点的运动速度与方向,提取有效特征点,得到人群运动目标图像;通过计算人群运动目标图像内光流点的方向熵、幅值熵与平均速度的乘积,确定运动混乱度;对比分析运动混乱度与设置阈值,完成人群异常行为检测。实验结果表明,该算法可有效提取人群视频帧内的角点和运动目标图像,准确检测人群异常行为,具有较好的视频监控中人群异常行为检测效果。

An explicit finite volume element method for solving characteristic level set equation on triangular grids

Level set methods are widely used for predicting evolutions of complex free surface topologies,such as the crystal and crack growth,bubbles and droplets deformation,spilling and breaking waves,and two-phase flow phenomena.This paper presents a characteristic level set equation which is derived from the two-dimensional level set equation by using the characteristic-based scheme.An explicit finite volume element method is developed to discretize the equation on triangular grids.Several examples are presented to demonstrate the performance of the proposed method for calculating interface evolutions in time.The proposed level set method is also coupled with the Navier-Stokes equations for two-phase immiscible incompressible flow analysis with surface tension.The Rayleigh-Taylor instability problem is used to test and evaluate the effectiveness of the proposed scheme.

STUDY OF KINETIC ANALYTICAL METHOD BASED ON CATALYTIC REACTIONS

In this thesis a reviwe is presented for the advances of catalytic spec trophotometric method, three new catalytic reaction systems were presented, and its rela live catalytic spectrophotometric methods have been established by flow injection tech nique. Particulary the relative chemometric methods and theories were used in the field of kinetic - catalytic analysis, in order to solve the problems of multicomponents analysis.

基于光流场分析的红外图像自动配准方法研究

提出了一种基于光流场分析的准确的红外图像自动配准方法 .该方法可分为两个过程 :先是利用全局光流场完成两幅图像背景区域的配准 ;其次利用由粗到细的层级匹配算法提取两幅图像中运动目标的特征点集 ,根据两组特征点集由最小二乘法计算出运动目标的变换参数 ,完成运动目标的配准 .对一定研究领域的红外图像自动配准的仿真实验表明 :该方法准确且对场景的运动有很好的鲁棒性 .

序列图像中运动目标检测与跟踪方法分析

运动目标检测与跟踪是计算机视觉研究和图像编码研究的主要内容,有着广泛的应用领域。研究了目前运动对象检测与跟踪的一些常用方法,包括时域差分法、背景差分法、基于光流场的检测方法和卡尔曼滤波、特征光流法的跟踪方法,并对各种方法进行了比较,指出其优缺点及适用范围,并给出了时域差分及背景差分方法的实验结果。

光流模值估计的微表情捕捉

采用力的加速度参量展开描述人脸表情的变化过程,直接反映变化速度,从而有效捕捉表情序列中由不完全肌肉运动所引起的微表情关键帧.利用Horn-Schunck(H-S)光流法对连续运动的人脸图像序列提取运动目标的运动特征;通过光学应变张量算法,结合运动特征中的光流速度估计,推导出加速度参量;利用全局阈值算法对加速度模值和速度与张量模值作分类、比较,实现微表情图像序列关键帧的提取.采用Oulu大学SMIC微表情数据库中16个实验对象的88个微表情片段作为实验样本,平均正确识别率可达80.7%,比仅利用光学张量算法的正确识别率高12.5%.实验结果表明,所提出的加速度参量对微表情提取更具有效性.

基于SURF与光流法的增强现实跟踪注册

提出一种基于SURF与光流法相结合的增强现实局部跟踪注册方法。采用光流法对移动对象区域进行跟踪,利用SURF算法仿射、尺度不变性及运算速度快的优点对该区域进行特征提取与匹配,利用相邻帧之间特征点的匹配关系求得三维注册矩阵,在保持注册精确性的同时降低了系统运算时间。实验结果表明该方法达到了实时跟踪与准确注册的效果,并且在环境变化时保持了较好的鲁棒性。

基于特征点检测与光流法的运动目标跟踪算法

为了解决当前运动目标跟踪算法在背景模型复杂和目标特征不明显的情况下,导致算法跟踪能力不足的问题,本文分别从特征点检测与光流法分析的角度出发,提出了基于特征点检测与光流法的运动目标跟踪算法。首先,根据图像梯度矩阵最小特征值,通过仿射变换,精确化特征点帧间匹配,排除伪特征点,达到精准检测运动目标特征点的目的。然后,基于图像像素守恒原理,进行2幅图像间变形评估,建立图像约束方程,进一步精确跟踪运动目标。最后,基于软件开发环境QTCreator实现算法,并系统集成。实验测试结果显示:与当前运动目标跟踪技术相比,本文算法拥有更高的准确性与稳定性。

LK光流法的多信息融合目标跟踪算法研究

针对改善动态目标物体跟踪过程中特征点选取的准确性及实时性问题,提出一种基于LK光流法的多信息融合目标跟踪方法并设计相关算法.将Sift提取出的特征点作为初始特征点,结合Harris角点检测函数R值,通过设定阈值,从中选取出具有高对比度的特征点作为目标特征点,对其进行光流法跟踪;然后引入在线学习检测器,使得在跟踪器跟踪失败之后可以更新目标位置,以实现稳定、准确跟踪.实验的结果从动态目标识别所需时间、特征点的数量、交并比以及每秒处理的帧数几个方面进行比较分析,证明所提算法能够实现准确跟踪,且实时性强.

基于高阶奇异值分解的驾驶疲劳识别

提出一种基于人脸运动特征和高阶奇异值分解的驾驶疲劳识别方法。利用光流技术计算人脸皮层的运动速度,以此作为疲劳特征。为消除身份、光照和姿态等因素对疲劳识别的影响,利用高阶奇异值分解将疲劳特征与身份信息、光照信息、姿态信息分离。在疲劳子空间采用余弦距离最近邻方法进行疲劳识别。对不同光照条件下、不同人、不同姿态的疲劳状态进行识别实验,实验结果表明,该方法具有较好的识别效果。

半稠密光流法的图像跟踪匹配算法

针对传统光流法图像特征缺失、边界及遮挡等处容易导致目标跟踪质量降低或丢失的问题,提出一种半稠密光流法实现图像特征的稳定跟踪。首先计算出图像中像素梯度变化较大的像素区域;其次利用时变图像灰度的时空梯度函数来计算像素的速度矢量,进而实现像素区域的跟踪;最后将状态向量作为剔除跟踪失败的依据,保留跟踪质量优良的像素区域。结果表明,该算法能有效地提高图像特征跟踪能力,保证图像中重要信息不丢失,同时保证运算速率。

基于互信息纠正的人群运动特征提取方法

针对光流法在估计人群运动速度时对噪声敏感的问题,提出一种基于互信息(Mutual Information,MI)纠正的人群运动特征提取方法。以稠密光流场为数据空间,在局部可重叠区域内计算各个速度向量与区域速度期望之间的无参数测度MI,通过直方图统计出MI偏小的速度向量,然后将这些向量向区域速度期望向量做出纠正处理,直至MI符合预设条件为止。实验表明,经MI纠正后可得到光滑的速度场,为后续分析人群行为提供增强的人群运动特征。

一种高效人体行为特征提取方法

针对现有人体行为识别方法中特征提取的计算负荷重的问题,提出一种新颖的基于光流的特征描述方法。由于背景的光流包含一定的噪声,提出几种方法来减弱噪声影响。首先,将固定数量的帧捆成一个块并提取光流的中值,其次,对基于整体量级的直方图进行正则化,最后在频率域进行低通滤波。利用傅立叶变换将时间域转化成频率域,保证时间变换时特征的不变性。在构造光流直方图时在直方图中分配行为的方向,因此可以保证行为表示时方向的不变性。最后,在KTH实验数据集与自建数据集中进行对比实验,结果表明该方法的有效性。

基于深度卷积特征光流的形变医学图像配准算法

光流法是一种基于光流场模型的重要而有效的形变配准算法。针对现有光流法所用特征质量不高使得配准结果不够准确的问题,将深度卷积神经网络特征和光流法相结合,提出了基于深度卷积特征光流(DCFOF)的形变医学图像配准算法。首先利用深度卷积神经网络稠密地提取图像中每个像素所在图像块的深度卷积特征,然后基于固定图像和浮动图像间的深度卷积特征差异求解光流场。通过提取图像的更为精确和鲁棒的深度学习特征,使求得的光流场更接近真实形变场,提升了配准精度。实验结果表明,所提算法能够更有效地解决形变医学图像配准问题,其配准精度优于Demons算法、尺度不变特征变换(SIFT)Flow算法以及医学图像专业配准软件Elastix。