基于小波变换和形态学的织物疵点边缘检测

为了精确确定织物疵点边缘,提出了一种基于小波变换和形态学的织物疵点边缘检测方法.在利用形态学实现疵点检测后,对其进行小波分解,用小波模极大值法和基于数学形态学的算法分别提取高低频子图像的疵点边缘,采用合理的融合规则将两个边缘图像进行融合.实验结果表明,该算法能有效地抑制噪声,且边缘清晰、准确,效果优于经典的边缘检测算法,具有可行性和有效性.     

基于高斯金字塔和视觉显著性的色织物疵点检测

目的为了提高色织物疵点检测的准确率。方法提出一种基于高斯金字塔和视觉显著性的色织物疵点检测方法。首先预处理待检测色织物图像,削弱不均匀光照和环境造成的影响;再对预处理后的图像进行灰度化,接着对灰度图进行高斯金字塔分层,然后对分层后的图像进行显著性处理,以获取图像的显著图;最后利用迭代阈值分割的方法对显著图进行阈值分割,得到色织物图像的疵点区域。结果将该方法与其他色织物疵点检测方法进行对比可知,检测的效果明显优于其他方法,疵点检测的准确率为91.25%。结论该方法可以有效地对色织物疵点进行检测,将疵点区域提取出来,为色织物后续生产中的加工处理提供有用信息。     

基于多片DSP+FPGA结构的织物自动检测系统设计与实现

应用四片DSP＋FPGA流水阵列结构，用现场可编程门阵列FPGA对采集的视频数字图像信息进行处理，实现了织物疵点自动检测。设计了以四片TMS320C62x为数字图像信息处理核心；由FPGA实现系统控制互连的织物疵点自动检测系统；设计了基于直方图统计和基于支撑矢量机的织物疵点分类识别算法。实验结果表明，该系统工作可靠、有效，能得到满意的检测结果。     

基于虚拟仪器的坯布疵点在线检测系统

为了能实时检测坯布产品质量,把坯布疵点缺陷标记出来并进行分类,本文研究了适用于坯布疵点检测的实时检测系统,论述了其硬件结构和软件流程．提出了一种基于虚拟仪器的快速有效的检测算法,对采集的图像进行实时分析、处理与识别．结果表明,该系统可以有效地检测出织物的疵点和疵点的类别．对于每帧4096×4096像素,实际大小为20cm×20cm的图像,检测速度最高可到达90m／min．     

基于小波分析的织物疵点视觉检测方法

采用了二维离散小波变换对织物图像进行处理,根据小波变换后高、低频分量的特点,提取分解子图像特征值,送入BP神经网络进行识别,从而实现织物疵点的检测.     

织物疵点检测与识别的方法研究

将织物图像分成大小相同的局部窗口,选取局部熵最小的窗口为待研究的感兴趣区域,在此区域内分割出疵点图像并用数学形态学中的开运算滤除噪声,计算疵点形状因子等作为识别参数,研究表明此方法因能避免对整幅图像进行复杂运算,在提取特征参数时对图像的全局搜索,具有识别正确率高、检测速度快等优点．     

液浸法超声检测的探头选择方法研究

针对使用液浸法超声检测时无行业标准指导其探头参数选择的问题,提出一种基于近场区长度的探头选择方法.该方法运用多元高斯模型计算介质中的声压分布,确定探头近场区长度,然后通过近场区与检测对象位置关系选择最佳探头参数.缺陷回波仿真实验结果表明:该探头选择方法是有效的,有助于提高液浸法超声检测的准确度.     

基于薄层单元法的土动力分析

推导了冲激荷载作用下薄层单元法的基本计算公式,运用这一方法对土中的波动传播进行了模拟,并给出了计算实例.该方法也可用于爆炸荷载作用下的土动力分析.     

基于熵方法广义关联度的织物检测

对织物的失真变形程度进行匹配检测.将熵方法广义关联度系数法的概念引入其中,以两组随机信号分别描述两幅织物图像的灰度特性,继而提出了一种以墒方法广义关联度系数为织物失真程度的指标.鉴于广义关联度系数是用于描述两个随机变量内在联系的,首先分析了广义关联度系数与两幅织物图像相似性的关系,其广义关联度系数的值愈大则图像愈相似,否则愈不相似,并进行了实验验证.实验结果表明,该算法是可行有效的.可以对织物图像的相似性进行定量描述.以界定织物失真变形的程度.     

超细径保偏光纤的对轴方法优化及实验

为了在光纤陀螺等关键器件中使用超细径保偏光纤取代常用的普通直径保偏光纤,需要对其偏振轴的确定方法进行改进和优化.首先,使用光线追迹法对图像侧面成像对轴法在几种条件下的理论图像进行计算机仿真,并对仿真结果较好的两种情形进行实验验证;然后,对实验数据曲线进行傅里叶展开以抑制随机干扰,将结果较好的三亮纹方法改进为三亮纹四点特征法,并对该方法进行实验验证;最后,测定使用三亮纹四点特征法得到的重复精度.实验结果表明,该方法大大提高了超细径保偏光纤偏振轴定轴过程中的抗干扰能力,重复精度达到1.6°,达到了普通直径保偏光纤的偏振轴定轴精度,满足实际应用要求.     

基于小波多尺度积与形态学的瑕疵检测算法

针对布匹瑕疵种类多以及单个方法仅对特定类瑕疵有效的问题,提出一种新的基于小波多尺度积和数学形态学的布匹瑕疵检测算法.首先对输入图像进行二进小波变换后,低频近似子图经过数学形态学运算,得到良好的瑕疵形状特征,然后对高频子图使用小波多尺度积方法,可以抑制噪声的同时增强瑕疵的边缘线性特征,最后融合得到最终的检测结果.实验结果...     

Layer modelling for the inspection allocation problem in re-entrant production systems

In the past two decades, re-entrant production systems have received a great deal of attention. One of the most important characteristics of re-entrant production systems is that the work units are manufactured layer-by-layer, so that some defects are more difficult to inspect after they are covered by the next layer. In this research, a mathematical model considering layered fabrication is developed to find the optimal solution for allocating inspections in re-entrant production systems. Workstations of variables data and inspections of quality characteristics measurement are modelled. In addition, this study considers three possibilities for the treatment of detected non-conforming units, namely repair, rework and scrap. Moreover, a heuristic algorithm is proposed in order to improve the optimization method based on complete enumeration, which suffers from a large amount of computation time, especially as the number of workstations increases. The model is highly extensible and applicable, so it can serve as a production-planning tool to solve the inspection allocation problem in re-entrant production systems.     

Methods for vibration diagnostics of the initial stages of damage of rotation systems

We present the main ideas of methods for early diagnostics of mechanical rotation systems based on the theory and statistics of periodically nonstationary random processes regarded as a mathematical model of signals of vibrations. New diagnostic criteria for defects are proposed, and new possibilities that they open are shown. The application of the developed methods to the early diagnostics and detection of defects of bearing units at the Dobrotvir thermal power plant is described.     

用于压/拉检测的机器人皮肤柔/弹性传感器阵列设计

基于炭黑/硅橡胶复合材料的压力以及拉伸敏感特性,设计了一种可用于检测机器人皮肤接触压力和拉伸特性的新型传感器阵列。该传感器阵列的弹性电极结构克服了传统传感器不可拉伸的缺点,实现了传感器阵列的柔/弹性;阵列中设计了9个检测压力的传感器单元和2个检测拉伸的传感器单元,通过传感器结构设计和拉/压干扰特性分析以及补偿算法解决了拉伸和压力同时测量时的干扰问题,并构建了求解压力与拉伸的数学模型。实验结果表明:该传感器阵列实现了对压力和拉伸的同步检测,可用于机器人柔性皮肤中关节等部位。     

基于增强的低对比度印品缺陷的识别技术研究

目的 针对当前印刷缺陷检测系统中存在的低对比度印刷缺陷检测精度不高等问题,基于HSV颜色空间,提出一种增强的低对比度印刷缺陷识别方法。方法 首先,将标准样张图像与采集到的印刷图像由RGB颜色空间转换到HSV颜色空间,并提取视觉上变化敏感的亮度分量V作为待检测对象;其次,将对比度受限的局部直方图均衡（CLAHE）与数学形态学相结合,来增强显现待检测图像中的缺陷;再次,使用连通域分析方法来获取缺陷的面积、周长、离心率、长宽比和圆形度等5种特征信息,并以此建立15个特征模型;最后,构建基于PNN的印刷缺陷识别神经网络,并在Matlab中实现对低对比度印刷缺陷的识别。结果 15个模型的平均耗时为475 ms,都控制在毫秒级别,满足了现代印刷缺陷检测对于实时性的要求。其中模型2的测试正确率为95%,能够识别污点等点缺陷,模型3和模型12的测试正确率为93%和93.3%,能够识别刮痕等线缺陷,模型5的测试正确率为93.1%,能够识别墨迹等面缺陷,且测试正确率高于基于BP神经网络的缺陷识别方法。结论 从缺陷检测的实时性和精确性上来讲,提出的方法能够对低对比度印刷缺陷进行实时和精确的检测。     

Ultrasonic nondestructive testing of scattering media using the decomposition of the time-reversal operator

In ultrasonic nondestructive testing, the iterative time-reversal process is an adaptive technique that can be used to detect flaws in complex samples with a large array of transducers. The decomposition of the time-reversal operator (DORT) method is a detection technique that is derived from the mathematical analysis of the iterative time-reversal process. Contrary to time-reversal techniques, the DORT method does not require programmable generators, and it allows the simultaneous detection and separation of several defects. In this paper, the method is applied to a Ti6-4 titanium cylindrical sample to separate the echo of a defect from the speckle due to microstructure contribution. The grain structure of this titanium alloy makes detection very difficult and, for large depths, conventional techniques do not allow the detection of small flaws with a satisfactory signal-to-noise ratio. The efficiency of the DORT method to detect a flat bottom hole with a diameter of 0.4 mm located at a depth of 140 mm in a titanium alloy sample is shown.     

A Novel Non-destructive Testing Method by Measuring the Change Rate of Magnetic Flux Leakage

The application of magnetic sensors in the traditional magnetic flux leakage (MFL) technique has a significant influence on the detection results. The sensor is typically used to directly measure the amplitude of the magnetic leakage flux intensity as the detection signal. In view of noise effects on the detection result and the subsequent misinterpretation of defect signals, a new non-destructive testing method is proposed. The proposed method intends to measure the magnetic flux change rate using two sensors. A mathematical model is first established to present the principle of the change rate measurement. Based on the magnetic dipole theory, it is inferred that the new method is applicable and sensitive to the detection and location of defects. Moreover, this method is advantageous as it inhibits the interference of MFL noises such as the distension noise, background noise, and vibration noise. The model predictions are then verified by a series of simulations. Finally, an experimental platform is set up to practically detect the defect of a steel plate, and the results agree with the demonstrations and simulations.     

Nanoconfinement of spider silk fibrils begets superior strength, extensibility, and toughness

Silk is an exceptionally strong, extensible, and tough material made from simple protein building blocks. The molecular structure of dragline spider silk repeat units consists of semiamorphous and nanocrystalline β-sheet protein domains. Here we show by a series of computational experiments how the nanoscale properties of silk repeat units are scaled up to create macroscopic silk fibers with outstanding mechanical properties despite the presence of cavities, tears, and cracks. We demonstrate that the geometric confinement of silk fibrils to diameters of 50 ± 30 nm is critical to facilitate a powerful mechanism by which hundreds of thousands of protein domains synergistically resist deformation and failure to provide enhanced strength, extensibility, and toughness at the macroscale, closely matching experimentally measured mechanical properties. Through this mechanism silk fibers exploit the full potential of the nanoscale building blocks, regardless of the details of microscopic loading conditions and despite the presence of large defects. Experimental results confirm that silk fibers are composed of silk fibril bundles with diameters in the range of 20-150 nm, in agreement with our predicted length scale. Our study reveals a general mechanism to map nanoscale properties to the macroscale and provides a potent design strategy toward novel fiber and bulk nanomaterials through hierarchical structures.     

非测地线缠绕稳定条件和缠绕工艺性的判别

本文首次给出一般曲面上非测地线缠绕稳定条件的推导, 推导结果还得到了该条件更精确的形式。为判别缠绕工艺性, 给出了一种确定曲面外法矢的算法, 以此为基础通过分析总结出了非测地线缠绕工艺性的判据。