基于数字图像处理技术的树脂眼镜镜片缺陷检测研究

通过设计的镜片自动检测光学硬件系统,给出了树脂镜片精细杂质图像处理措施,将图像差分法应用于镜片图像边缘的去除;同时解决了镜片检测灰尘影响的问题;另外运用改进型的归一法解决杂质计数问题,为自动化分级分拣系统的研发提供重要的数据支持,为镜片缺陷检测系统的后续分级分拣的数字控制模块提供了可操作性。该检测系统检测精度为0.03 mm、速度为1片/s,检测误差控制在5%。     

一种眼镜镜片缺陷自动化检测系统的研究

根据光散射原理和视觉检测的方法,研究开发了一种基于机器视觉的眼镜镜片缺陷自动化检测系统,对眼镜镜片进行图像获取、简单的图像处理与分级、分拣,实现对镜片的分类。系统采用并行机构运行,利用低角度前向照明方式获得了镜片表面和内部的斑、杂质、羽毛等缺陷图像;利用归一算法进行图像处理与分级,得到不同缺陷的识别信息。实验结果表明,系统能够检测出镜片中含有的所有缺陷,每个镜片的平均检测时间控制在2 s以内,检测精度为0.038 mm。     

A study on coherent structures and drag-reduction in the wall turbulence with polymer additives by TRPIV

An experimental measurement was performed using time-resolved particle image velocimetry (TRPIV) to investigate the spatial topological character of coherent structures in wall-bounded turbulence of polymer additive solution. The fully developed near-wall turbulent flow fields with and without polymer additives at the same Reynolds number were measured by TRPIV in a water channel. The comparisons of turbulent statistics confirm that due to viscoelastic structure of long-chain polymers, the wall-normal velocity fluctuation and Reynolds shear stress in the near-wall region are suppressed significantly. Furthermore, it is noted that such a behavior of polymers is closely related to the decease of the motion of the second and forth quadrants, i.e., the ejection and sweep events, in the near-wall region. The spatial topological mode of coherent structures during bursts has been extracted by the new mu-level criteria based on locally averaged velocity structure function. Although the general shapes of coherent structures are unchanged by polymer additives, the fluctuating velocity, velocity gradient, velocity strain rate and vorticity of coherent structures during burst events are suppressed in the polymer additive solution compared with that in water. The results show that due to the polymer additives the occurrence and intensity of coherent structures are suppressed, leading to drag reduction.     

Superelastic shape memory alloy cables: Part II – Subcomponent isothermal responses

This paper constitutes the second part of our experimental study of the thermo-mechanical behavior of superelastic NiTi shape memory alloy cables. Part I introduced the fundamental, room temperature, tensile responses of two cable designs (7 × 7 right regular lay, and 1 × 27 alternating lay). In Part II, each cable behavior is studied further by breaking down the response into the contributions of its hierarchical subcomponents. Selected wire strands were extracted from the two cable constructions, and their quasi-static tension responses were measured using the same experimental setup of Part I. Consistent with the shallow wire helix angles in the 7 × 7 construction, the force–elongation responses of the core wire, 1 × 7 core strand and full 7 × 7 cable were similar on a normalized basis, with only a slight decrease in transformation force plateaus and slight increase in plateau strains in this specimen sequence. By contrast, each successive 1 × 27 component (1 × 6 core strand, 1 × 15 strand, and full cable) included an additional outer layer of wires with a larger number of wires, greater helix radius, and deeper helix angle, so the normalized axial load responses became significantly more compliant. Each specimen in the sequence also exhibited progressively larger strains at failure, reaching 40% strain in the full 1 × 27 cable.Stress-induced phase transformations involved localized strain/temperature and front propagation in all of the tested 7 × 7 components but none of the 1 × 27 components aside from the 1 × 27 core wire. Stereo digital image correlation measurements revealed finer features within a global transformation front of the 1 × 7 core strand than the 7 × 7 cable, consisting of an staggered pattern of individual wire fronts that moved in lock-step during elongation. Although the 1 × 27 multi-layer strands exhibited temperature/strain localizations in a distributed pattern during transformations, the localizations did not propagate and their cause was traced back to contact indentations (stress concentrations) arising from the cable’s fabrication. The normalized axial torque responses of the multi-layer 1 × 27 components during transformation were distinctly non-monotonic and complex, due to the alternating handedness of the layers. Force and torque contributions of individual wire layers were deduced by subtracting 1 × 27 component responses, which helped to clarify the transformation kinetics within each layer and explain the unusual force and torque undulations seen in the 1 × 27 cable response of Part I.     

风吹雪廓线的风洞实验研究

利用颗粒图像测速仪对新雪形成的风吹雪进行风洞实验研究, 给出了不同高度处雪粒粒径分布函数以及平均粒径廓线、雪粒数通量廓线的分布规律. 发现当摩阻风速大于0.5 m/s 时单宽输雪率与摩阻风速满足指数函数的关系, 小于0.5 m/s 时两者满足幂函数的关系, 总体而言, 单宽输雪率与摩阻风速呈线性关系.     

Self-consistency and the use of correlated stochastic separated flow models for prediction of particle-laden flows

Computational fluid dynamics (CFD) is being used increasingly in the design and analysis of particle-laden flows. A significant challenge of this work is in correctly predicting the interaction of the fluid turbulence with the particulate phase. Typically, Lagrangian tracking is used to calculate the particle trajectories with stochastic treatments used to provide an instantaneous turbulent flow field. The stochastic calculations are based on the mean velocities and turbulence quantities calculated by the CFD solver. The current work examines the correlated stochastic separated flow (SSF) model used to synthesize the instantaneous fluid velocity field. Two functional forms of the Eulerian spatial correlation are considered: exponential, and Frenkiel with loop parameter m equal to unity. It is well known that the use of a Frenkiel function is incorrect due to the Markovian nature of the model. Nonetheless, a literature review indicates that the Frenkiel function is still being used in the CFD community. In order to illustrate the implications of this, numerical predictions are compared to Taylor's analytical result for fluid particle dispersion in homogeneous isotropic turbulence. Excellent predictions are obtained with the exponential correlation and recommendations on timestep requirements are made. In contrast, predictions from the Frenkiel model are in poor agreement with Taylor's solution. This poor agreement results from an inconsistency between the effective correlation of fluid velocities arising from the model and the original intended correlation.     

基于干涉原理的双图像加密系统

提出一种新的虚拟光学加密方法,该方法分3个步骤将两幅图像隐藏于3个随机相位板中。将两幅图像分别作为相位和幅度信息隐藏于一个复数场中,利用随机相位板作为参考光,并采用数字全息术将该复数场转换成平稳随机白噪声,利用干涉加密原理进一步将此白噪声储存于2个随机相位板中,从而将两幅图像信息隐藏于3个纯随机相位板中。该方法原理简单,无需进行迭代操作。计算机模拟结果显示:任何一个密钥错误,均无法得到原始图像的任何信息。同时,解密过程对系统中距离及波长等参数非常敏感,较大地提升了系统的安全性。     

原位成形光学微透镜的聚光性能研究

为解决荧光谱微检测系统中光学微透镜制作的困难,提出一种原位成形微透镜的制作方法,研究该微透镜的聚光性能,利用传统几何光学原理,分别推导出激发微透镜与检测微透镜的光线追迹方程,用该方程计算了2种微透镜的照射点分布几率,画出相应的分布曲线图,讨论了不同形状参数的微透镜对聚光性能的影响。研究结果表明,通过在光学胶中适当掺杂及打磨基底玻璃,可减小接触角,加大微透镜的形状参数,既能坚固透镜与基底间的粘接强度,又能提高微透镜的聚光本领。制作了使用原位成形光学微透镜的荧光谱微检测器,对生物荧光试剂进行了测试。实验结果表明,使用原位成形光学微透镜进行聚光,可使生物荧光谱强度提高4倍以上,通过在光学胶中掺杂以及打磨玻璃的办法,可以提高透镜的聚光本领约1.2倍。     

基于奇异值分解的运动模糊图像广义逆恢复方法

当线性模型应用于运动模糊模糊图像的恢复时,方程的最小二乘解是恢复图像的最优线性无偏估计。由于图像退化过程的不适定性,当观测值受到噪声干扰时,该解往往会远偏离真值。为了克服这个问题,通过对退化矩阵的奇异值分解,提取其不易受干扰的子空间,用该空间重构的逆矩阵具有良好抑噪能力,使图像在较长的运动模糊尺度内恢复时保持较低的失真。     

基于峰值信噪比和小波方向特性的图像奇异值去噪技术

提出一种利用小波变换子图像不同的方向特性和峰值信噪比进行奇异值分解的图像去噪算法。由于图像经过小波变换后,低频子图像集中了原图像的大部分能量噪声,故仅作简单维纳滤波;而噪声则主要集中在小波域中的三个不同方向的高频子图中,且系数较小,因此可以利用奇异值分解进行去噪处理,即用较大的奇异值和对应的特征向量重构出去噪图像,然而由于奇异值分解固有的行列方向性,对于高频对角线子图重构出的图像去噪效果不理想,故采取旋转至行列方向后再进行常用的奇异值滤波;最后将去噪后的低频和高频子图进行小波反变换重构出最终的去噪图像,其中重构所需的奇异值个数由图像的峰值信噪比确定。 实验结果表明,该方法在有效去噪的同时较好的保留了原有的高频细节信息。