微弱棉花异性纤维图像的多通道小波分割方法研究EI北大核心CSCD

针对皮棉异性纤维快速检测过程中产生的微弱异性纤维目标,提出一种基于多通道色彩分量小波分解、重构的异性纤维分割方法。将原始异性纤维图像分解成3幅单通道图像,分别对每幅单通道图像进行小波分解,并对小波分解后的3幅小波图像进行两两做差处理,将做差处理后得到的图像进行小波逆变换图像灰度重构。试验结果表明,该方法能将微弱异性纤维目标准确地分割出来,解决现有分割方法无法直接分割的难题。     

微弱棉花异性纤维图像的多通道小波分割方法研究

针对皮棉异性纤维快速检测过程中产生的微弱异性纤维目标,提出一种基于多通道色彩分量小波分解、重构的异性纤维分割方法。将原始异性纤维图像分解成3幅单通道图像,分别对每幅单通道图像进行小波分解,并对小波分解后的3幅小波图像进行两两做差处理,将做差处理后得到的图像进行小波逆变换图像灰度重构。试验结果表明,该方法能将微弱异性纤维目标准确地分割出来,解决现有分割方法无法直接分割的难题。     

棉花异性纤维分拣机器人抓棉手的结构设计

棉花中的异性纤维影响纯棉产品质量，采用智能机器人替代人工作业是新的发展方向。文章设计了一款具有5自由度棉花异性纤维分拣机器人。该机器人主要由四轮移动小车、液压升降平台、腰部、大臂、小臂、机械手等组成。针对棉花抓取与分拣，对机器人的抓棉手进行了结构设计，以滑环和连杆作为设计变量，优化了抓棉手的机械尺寸；同时，文章针对抓棉手指采用有限元方法进行了应变和模态分析，结果显示手指设计合理。     

棉花中异性纤维识别方法的研究

皮棉异性纤维的识别是皮棉自动剔除系统的关键环节,文中通过分析直方图,选择了合适的图像增强方法,改进了Otsu分割算法,并提高了计算速度.实验结果表明,所选择的图像增强方法和改进的Otsu算法能够有效地提高图像分割的质量.     

基于水平特征的车辆检测与跟踪

利用加权平均滤波算法对图像进行了滤波处理,去除部分噪声,利用Sobel算子的水平算子对图像的水平边缘进行增强处理,利用最大类间方差法对边缘增强后的图像进行图像分割;提出了利用车辆的水平特征中的列直方图和行直方图对车辆进行检测,根据检测结果得出该方法检测效果良好,即使是复杂环境下的车辆也能得到有效的识别,且对货车的识别效果良好;最后,利用限定感兴趣区域法对车辆进行跟踪,验证了识别方法与跟踪方法的有效性。     

基于小波多尺度分解的区域图像融合及算法研究

提出一种新的图像融合方法,该方法首先利用小波变换对要融合的图像进行多尺度小波分解;然后对分解后的各层上不同频带的子图像采用不同的融合处理技术,即在低频城内采用平均算子进行融合处理,以保留图像的背景信息,在高频域内对分解后的小波系数矩阵采用区域特征进行融合处理;最后将融合的低频分量和高频细节分量结合进行小波逆变换得到融合图像.采用该融合算法对两幅不同位置聚焦的图像进行融合实验,结果表明,采用该方法可以得到较好的融合效果.     

数学形态学和小波变换的红外图像处理方法

为了从红外热图像中识别和跟踪目标,提出了一种采用数学形态学和小波变换相结合的红外图像处理新方法。该方法首先利用形态开、闭滤波器组对原始图像进行平滑处理;然后采用一种改进的数学形态学的分水岭算法对红外图像进行分割,同时利用考虑了图像纹理信息的小波阈值对分割的图像进行滤波处理;最后针对过分割问题提出了一种利用区域灰度中值对分割区域进行融合处理的算法。实验结果及算法性能评估结果表明,该方法能够较好地解决红外目标的识别问题,具有较好的实用性。     

基于相关匹配的QR码识别方法

QR码图像识别是QR码应用中的关键技术.Hough变换、曲面拟合去背景、控制点变换等方法为QR码识别过程中图像预处理的基本方法,针对图像预处理后图像识别率低的缺点,提出了一种基于相关匹配法的QR码识别方法.基于曲面拟合的改进自适应阈值法分割图像.采用Hough变换和控制点变换方法校正图像的几何失真变形.利用模板对QR码进行相关匹配,对相干系数阈值处理得出取样网格.实验表明,本文算法能有效提高QR码识别效率和效果.     

基于OpenCV图像处理的车道线识别研究

近年来,智能驾驶技术成为国内外学者研究的热门,车道线识别技术作为智能驾驶领域的一项关键技术,需要准确地识别出各种条件下的车道线。本文设计了一种基于OpenCV图像处理的车道线识别算法,该算法首先将图像进行色彩空间转换,然后利用大津法（OTSU）和Canny算子对图像进行分割,设计掩膜找出图像中的感兴趣区域,采用渐近概率式Hough变换对直线车道线进行检测,拟合直线绘制出车道线,最后借助OpenCV平台对算法进行实现,结果表明该算法对直线车道线识别具有很好的效果。     

人民币纸币识别系统

设计研究人民币纸币清分系统中的主要模块,如图像采集、处理和识别.在硬件部分,选用CM326A4作为系统的图像传感器,选用DSP芯片TMS320C6711进行纸币图像的处理和识别;在软件部分,选取预处理之后的纸币图像的尺寸特征用模糊逻辑推理方法识别图像的面值,提取矩作为特征,采用三层BP神经网络来识别纸币的正反面和正反向,并通过实验证明该系统的有效性和可行性.     

基于CPU＋GPU异构系统的棉花异纤识别系统

棉花中的异性纤维给棉纺织企业带来了不小的损失,因此开发出高效率的异纤分拣机是很多纺织企业的迫切需求。现有的异纤分拣机虽然能够达到剔除部分异纤的效果,但由于图像处理运算都是在工控机上的CPU里实现,为了达到在线实时检测的需求,只能采用比较简单的识别算法,异纤识别效果不是很理想。本文通过研究基于CPU＋GPU异构系统的棉花异纤识别系统,让GPU通过并行运算来实现异纤的识别算法,从而大大提高了运算效率,有效的减少了算法的运行时间,为复杂的异纤识别算法在异纤识别系统的使用提供了条件,从而有效提高异纤杂质的剔除率。实验表明,基于CPU＋GPU异构系统对算法的时间提高了十几倍。     

光纤端面参数自动化测量系统的研究

为快速、准确、方便地进行光纤端面几何参数的测量,在数字图像采集和处理的基础上,研究和设计了光纤端面参数自动化测量系统。介绍了采用图像处理方式进行光纤端面几何参数自动测量的基本原理。对比了Canny边缘检测法和高通滤波法对光纤图像进行边缘检测的效果,体现出高通滤波在光纤端面图像处理中的优势。针对普通光纤、保偏光纤和光子晶体光纤不同的端面形状,设计出利用图像处理方法自动测量各种光纤端面几何参数的软件系统。结果表明该系统可有效地测量出各类光纤端面的几何参数。     

棉纤维长度分布测量仪器的开发

本文提出了一种新的棉纤维综合长度指标测量方法，该方法的特点是首先测量出纤维试样的长度分布，然后计算各个长度指标。它的技术创新点有2个：（1）可测Hertel取样；（2）图像测量。由于涨落的存在，实际Hertel取样结果与理论Hertel取样有偏差。为此，通过使用中值滤波模板的处理，保证了两者的高度吻合。一个近显微光学测量系统，保证了在照影仪曲线上任意一点，棉纤维试样根数的测量误差不超过3％。与Y111的对比实验表明，主体长度、短纤维率和品质长度这3个指标，相关系数的平方R^2分别为0．94、0．97和0．82。该仪器对一个棉纤维试样的测量能够在15min内完成。     

Characteristics and formation mechanism for stainless steel fiber with periodic micro-fins

Metal fibers have been widely used in many industrial applications due to their unique advantages. In certain applications, such as catalyst supports or orthopedic implants, a rough surface or tiny outshoots on the surface of metal fibers to increase surface area are needed. However, it has not been concerned about the surface morphologies of metal fiber in the current research of metal fiber manufacturing. In this paper, a special multi-tooth tool composed of a row of triangular tiny teeth is designed. The entire cutting layer of multi-tooth tool bifurcates into several thin cutting layers due to tiny teeth involved in cutting. As a result, several stainless steel fibers with periodic micro-fins are produced simultaneously. Morphology of periodic micro-fins is found to be diverse and can be classified into three categories: unilateral plane, unilateral tapering and bilateral. There are two forming mechanisms for the micro-fins. One is that periodic burrs remained on the free side of cutting layer of a tiny tooth create micro-fins of stainless steel fiber produced by the next neighboring tiny tooth; the other is that the connections between two fibers stuck together come to be micro-fins if the two fibers are finally detached. Influence of cutting conditions on formation of micro-fins is investigated. Experimental results show that cutting depth has no significant effect on micro-fin formation, high cutting speed is conducive to micro-fin formation, and feed should be between 0.12 mm/r and 0.2 mm/r to reliably obtain stainless steel fiber with micro-fins. This research presents a new pattern of stainless steel fiber characterized by periodic micro-fins formed on the edge of fiber and its manufacturing method.     

线阵相机场曲复原及在异纤检测中的应用

为了消除相机的场曲给成像带来的空变模糊,提高图像分辨率和检测精度,针对在工业检测中广泛使用的线阵相机,提出了一种消除场曲影响的一维图像复原方法。在分析了空变模糊矩阵结构的基础上,通过估算部分区域的点扩散函数,再由偏移和插值得到全像场范围内的模糊矩阵。由模糊矩阵利用约束最小二乘法得到与观测信号无关的复原矩阵。检测过程中,将采集到的图像信号与复原矩阵相乘,获得复原后的图像。采用检测棉流内异性纤维的异纤检测系统,用含有较细异纤的实际数据对方法进行验证。结果表明,本方法在提高图像边缘处分辨率的同时增强了异常点与背景的差异,异常比增加了10%以上,为进一步提高检测精度创造了条件。     

On the processing and morphological aspects of metal fibers based on low-speed multi-tooth dry cutting

This study presents an investigation on both the processing and morphological aspects of cutting-based metal fibers. A self-designed multi-tooth cutter is used to efficiently make continuous long fibers under a low-speed dry-cutting condition. A finite element method-aided analysis is conducted to elucidate the mechanisms of this method. Based on both simulated and experimental results, it is found that fiber morphology shows different characteristics when different processing parameters are used. Moreover, the fiber structure and appearance also significantly depends on the mechanical properties of raw materials. It is shown that the Cu fiber belongs to flow chip due to highly plastic deformation, while the hardened Cu–Be, Cu–Be–Ni alloy and SUS316L fibers exhibit fractured features. The combined segmented vertical fins and sawtooth structures can be periodically formed on the Cu–Be fibers under an appropriate condition. Besides these findings, the effects of sintering condition on the fiber morphology are also reported in this work.     

基于BP神经网络的电缆沟监控模式识别方法

为了减少电缆沟盲目施工与偷盗等因素造成的损失和影响,引入一种智能的线路监控技术。该技术基于分布式光纤传感技术,利用BP神经网络对光纤传感器的四种常见的扰动信号进行模式识别。通过将光纤传感器得到的时域扰动信号用特定的程序处理后转换成图像,再经特定的图像处理流程后形成模式识别的样本。利用这些样本训练BP神经网络,并将训练好的模型应用到实际的电缆沟安全监控系统中进行测试。测试结果表明,电缆沟的总体识别成功率为98.16%,该识别方法还可以应用于光纤周界安防系统等领域。     

Application of the technique of environmental scanning electron microscopy to the paper industry

The environmental scanning electron microscope was used in conjunction with a heating stage to study in situ the polymerization process in the manufacturing of filter papers for the automotive industry. The images obtained from the video recording were converted into TIFF format with a computer program and were then analyzed using an image analyzer to assess the percentage shrinkage during the cure process. These experiments yield valuable information that is impossible to obtain with other electron microscopy techniques, for example, determining whether the polymer envelops the fibers or not, or whether there is a greater or lesser degree of anchoring between fibers.     

Preparing and analyzing fractured archaeological fibers

A technique was developed to prepare archaeological fiber cross sections for electron microscopic examination and x-ray analysis. Use of this new method allows chemical and morphological information to be obtained from the interior of a single fiber or yarn. Fibers are fractured while frozen and then freeze dried. Following mounting and carbon coating, fibers are examined by scanning and backscatter electron microscopy and then analyzed by using energydispersive spectrometry. Elemental distribution is mapped by using image-processing software. In this report, the described technique is employed in the examination of ancient fibers from three different long-term storage environments (moist buried, dry buried, museum stored). Data obtained by examining the interior of fibers such as these provide insight into the conditions of a fiber's growth, the treatments applied during the fiber's processing and use, and the conditions in which the fiber was stored.