Simultaneous determination of the optimal production–inventory and product inspection policies for a deteriorating production system

In this paper, we attempt to find a method for the optimization of production–inventory and product inspection policies for deteriorating production systems. Taking advantage of the nature of a deteriorating production system, a strategy would be not to inspect the first s items of the batch. Therefore, an inspection policy which disregards the first s (DTF-s) items of the batch is proposed. Under the DTF-s policy, we do not inspect the first s produced items but inspect only those items from the (s+1)th till the end of the production run. The objective of this study was the joint determination of the production lot size and the inspection policy s, resulting in a minimization of the expected average cost per unit time. Based on this model, the underlying conditions necessary for the existence of an optimal policy are given. Two commonly used inspection strategies, no inspection and full inspection are discussed. Under both inspection strategies, an optimal production–inventory lot is bounded by the traditional economic quantity. The case of full inspection is shown to be an extension of previously reported results. The option of investing in the process of quality improvement is also discussed. Finally, numerical examples are given to illustrate the method and its advantages in the conclusion.Scope and purposeThis paper considers the relationship between production, inventory and inspection in a deteriorating production system which may transit from the “in-control” state to the “out-of-control” state after a period of operation. Once the transition to the “out-of-control” state has occurred, it is assumed that some percentage of the items produced are defective or of substandard quality. However, in many cases, defects in a defective item can only be identified by an inspection process which carries an inspection cost. Those inspected items which are found to be defective are reworked at some cost before being shipped. On the other hand, defective items which are not inspected will be passed to the customer, incurring a much larger warranty cost. In order to operate such a system economically, tradeoffs among production setup, inventory, inspection and defective cost must be analyzed. Deterioration of the production system is an inherent process in all manufacturing industries. An understanding of the relationship among production, inventory and inspection for such systems will help managers to maintain efficient and economic control of operations.     

A comparison between the first absolute central moment and the gradient of Gaussian when detecting edges in presence of Gaussian noise

The gray-level absolute central moment of the first order provides ridges at gray-level discontinuities as well as a conventional gradient of Gaussian (GoG). A mass center b of the gray-level variability can also be associated to the first absolute central moment. When given a starting point p, vector b indicates the path which joins p to the nearest gray-level discontinuity as well as the gradient of the GoG magnitude. However, when the right configuration of the operator is chosen, vector b indicates a point which is closer to the discontinuity than p, regardless of the distance between p and the discontinuity. Therefore, when using vector b, gray-level discontinuities can be located with an iterative approach. In this paper, the edge detection properties of the first absolute central moment are compared with those of the GoG. The text was submitted by the authors in English. Marcello Demi was born in Cecina, Italy, in 1956. He graduated in electronic engineering from the University of Florence, Italy, in 1985. He is currently head of the Computer Vision Group at the CNR Institute of Clinical Physiology in Pisa, senior scientist for ESAOTE Spa, and he teaches a course on medical image processing at the Faculty of Applied Physics, University of Pisa. His research interests are cardiovascular image processing systems, contour tracking of moving deformable objects, and filtering schemes inspired by the early stages of biological vision systems. He has 80 scientific publications and his objective is the development of common projects with people who work in the area of biological vision for the purpose of both understanding biological vision and developing image processing systems. Francesco Faita was born on June 28, 1973, in La Spezia (Italy). In 2001, he graduated from Universitá degli Studi di Pisa obtaining the degree of electronic engineer. Since 2001, he has been working as a research fellow at the Institute of Clinical Physiology of the Italian National Research Council. His main research interests lie in computer vision, in particular, in the field of ultrasound imaging. Elisabetta Bianchini was born in Lucca, Italy, in 1975. She received her degree in electronic engineering from the University of Pisa, Italy, in 2004. Since 2004, she has been a junior researcher at CNR, the Italian National Research Council, at the DSP lab in IFC (Institute of Clinical Physiology). Vincenzo Gemignani was born on October 10, 1969, in Viareggio (Italy). In 1995, he graduated from Universitá degli Studi di Pisa obtaining the degree of electronic engineer. Since 1996, he has been working as a research fellow at the Institute of Clinical Physiology of the Italian National Research Council. His main research interests lie in digital signal processing, in particular, in field real-time image analysis.     

联合矩阵F范数的低秩图像去噪

摘 要：目的：低秩矩阵恢复是通过最小化矩阵核范数来获得低秩解,然而待恢复低秩矩阵相关性低的要求往往会导致求解不稳定的情况。方法：针对该问题,研究一种基于变量分裂的低秩图像恢复去噪算法,引入待恢复矩阵的Frobenius范数作为新正则项,与原有低秩矩阵的核范数组成联合正则化项,对问题进行凸松弛后,采用变量分裂的增广拉格朗日乘子法求解。结果：为考察方法的稳定性和去噪能力,选取了不同参数类型的加噪图像进行仿真,并结合恢复时间、信噪比、差错率等评价标准与现有低秩矩阵恢复算法进行对比。结论：实验结果表明增加Frobenius范数的低秩矩阵恢复模型在保持原有低秩稀疏恢复的前提下,具有良好的去噪性能,对相关性强的低秩图像恢复结果稳定性好,获得了更高的信噪比。     

一种基于计算机视觉的织物疵点自动检测方法

提出了一种基于计算机视觉的织物疵点的自动检测与识别方法,使用CCD摄像头对织物进行实时拍摄,对所得图片通过图像处理、小波分析,提取特征参数,有效地检测与识别缺经、缺纬、油污、破洞等常见疵点,其具有检测速度快、识别正确率较高等优点。     

基于机器视觉的二次设备运行状态巡检方法研究

为了提高二次设备运行状态的自动检测能力,提出一种基于机器视觉的二次设备运行状态巡检方法,采用机器视觉方法构建二次设备运行状态的成像模型,提取二次设备运行状态的视觉成像特征量,采用边缘轮廓检测方法进行二次设备运行状态机器视觉的边界特征分析,在机器视觉的梯度下降流中构建二次设备运行状态巡检的变结构分布模型,以灰度值较高区域边界像素点为中心,进行二次设备运行状态巡检的视觉重构识别,实现二次设备运行状态自动巡检。仿真结果表明,采用该方法进行二次设备运行状态巡检的智能性较高,特征分辨能力较好,提高了二次设备的巡检能力。     

光场成像技术及其在计算机视觉中的应用

目的 光场成像技术刚刚在计算机视觉研究中展开初步应用,其相关研究比较零散,缺乏系统性。本文旨在系统介绍光场成像技术发展以及其应用在计算机视觉研究中有代表性的工作。方法 从解决计算机视觉问题的角度出发,4个层面讨论光场成像技术最近十年的研究工作,包括:1)主流的光场成像设备及其作为计算机视觉传感器的优点与不足;2)光场相机作为视觉传感器的标定、解码以及预处理方法;3)基于4维光场的图像渲染与重建技术,以及其如何促进计算机视觉研究;4)以4维光场数据为基础的特征表达方法。结果 逐层梳理出光场成像在求解视觉问题中的优势和局限,分析其中根本性的原理与掣肘,力图总结出亟待解决的关键问题以及未来的发展趋势。结论 作为一种颇具前景的新型计算机视觉传感器技术,光场成像技术的研究必将更为广泛和深入。研究应用于计算机视觉的光场成像技术将有力的引导和促进计算机视觉和光场成像技术协同发展。     

双目系统全视域采样的SVM标定模型分析

目的 为改善摄像机间接标定采样不全、模型表达模糊问题,实现小视场下检测视域完备采样,提出一种基于双目系统全视域采样的支持向量机（SVM）标定方法。方法 该方法利用六角晶格标定板靶点序号可读特点为基础,采集整个双目系统有效视域中检测点的视差坐标、世界坐标并建立完备的样本集。选取SVM对该样本集进行训练,将SVM算法得到的模型参数代入其决策函数中进行求解,获得公式化的标定模型。由于六角晶格标定板的四角和中心分布了5个互为非中心对称的多边形,可在标定板部分区域被采集的情况下获取标定板位姿信息,进而读取采集的各靶点序号。通过上下移动标定板,利用HALCON算子获取图像中各靶点的序号,建立双目视觉系统检测区域的完备样本集。最后,利用SVM算法训练样本获得标定模型,可以明确表达出标定模型的数学形式。结果 与传统采样建立的模型进行对比分析,实验结果表明该方法建立模型的标定误差减小了24.51%,降低了标定模型在传统方法未采样区域的标定误差,证明了该方法的可行性。结论 提出一种基于双目系统全视域采样的支持向量机标定方法,通过非中心对称的多边形确定标定板上靶点的序号,实现双目视觉系统检测视域的完备采样。实验结果表明该方法提高了摄像机间接标定的精度,具有良好的适用性和鲁棒性,适用于小视域内双目视觉系统的间接标定。     

自适应K-means聚类的散乱点云精简

目的 点云精简是曲面重建等点云处理的一个重要前提,针对以往散乱点云精简算法的精简结果存在失真较大、空洞及不适用于片状点云的问题,提出一种自适应K-means聚类的点云精简算法。方法 首先,根据k邻域计算每个数据点的曲率、点法向与邻域点法向夹角的平均值、点到邻域重心的距离、点到邻域点的平均距离,据此运用多判别参数混合的特征提取方法识别并保留特征点,包括曲面尖锐点和边界点;然后,对点云数据建立自适应八叉树,为K-means聚类提供与点云密度分布相关的初始化聚类中心以及K值;最后,遍历整个聚类,如果聚类结果中含有特征点则剔除其中的特征点并更新聚类中心,计算更新后聚类中数据点的最大曲率差,将最大曲率差大于设定阈值的聚类进行细分,保留最终聚类中距聚类中心最近的数据点。结果 在聚类方面,将传统的K-means聚类和自适应K-means聚类算法应用于bunny点云,后者在聚类的迭代次数、评价函数值和时间上均优于前者;在精简方面,将提出的精简算法应用于封闭及片状两种不同类型的点云,在精简比例为1/5时fandisk及saddle模型的精简误差分别为0.29×10^-3、-0.41×10^-3和0.037、-0.094,对于片状的saddle点云模型,其边界收缩误差为0.030 805,均小于栅格法和曲率法。结论 本文提出的散乱点云精简算法可应用于封闭及片状点云,精简后的数据点分布均匀无空洞,对片状点云进行精简时能够保护模型的边界数据点。     

Filling n-sided regions with G 1 triangular Coons B-spline patches

Filling n-sided regions is an essential operation in shape and surface modeling. Positional and tangential continuities are highly required in designing and manufacturing. We propose a method for filling n-sided regions with untrimmed triangular Coons B-spline patches, preserving G ¹ continuity exactly. The algorithm first computes a central point, a central normal, the central, and the corner derivative vectors. Then the region is split into n triangular areas by connecting the central point to each corner of the boundary. These inner curves and all cross-boundary derivatives are computed fulfilling G ¹ compatibility conditions. And finally, the triangular patches are generated in the Coons B-spline form, one boundary of which is regressed to the central vertex. Neither positional nor tangential error is introduced by this method. And only one degree elevation is needed.     

Segmentation of stick text based on sub connected area analysis

A new stick text segmentation method based on the sub connected area analysis is introduced in this paper.The foundation of this method is the sub connected area representation of text image that can represent all connected areas in an image efficiently.This method consists mainly of four steps:sub connected area classification,finding initial boundary following point,finding optimal segmentation point by boundary tracing,and text segmentaton.This method is similar to boundary analysis method but is more efficient than boundary analysis.     

Development of a volumetric projection technique for the digital evaluation of field of view

Current regulations for field of view requirements in road vehicles are defined by 2D areas projected on the ground plane. This paper discusses the development of a new software-based volumetric field of view projection tool and its implementation within an existing digital human modelling system. In addition, the exploitation of this new tool is highlighted through its use in a UK Department for Transport funded research project exploring the current concerns with driver vision. Focusing specifically on rearwards visibility in small and medium passenger vehicles, the volumetric approach is shown to provide a number of distinct advantages. The ability to explore multiple projections of both direct vision (through windows) and indirect vision (through mirrors) provides a greater understanding of the field of view environment afforded to the driver whilst still maintaining compatibility with the 2D projections of the regulatory standards.

Practitioner Summary: Field of view requirements for drivers of road vehicles are defined by simplified 2D areas projected onto the ground plane. However, driver vision is a complex 3D problem. This paper presents the development of a new software-based 3D volumetric projection technique and its implementation in the evaluation of driver vision in small- and medium-sized passenger vehicles.     

自主移动机器人自适应室外道路检测

目的 为降低室外自主移动机器人视觉导航中遇到的阴影、裂纹及道路边界不规则造成的道路检测算法不鲁棒性,提出一种每帧灰度阈值可调的快速自适应道路检测方法。方法 先采用2维离散小波进行道路图像分解与重构,比较各级小波重构后的近似道路图像,确定出不影响“路-非路”灰度二分类的最佳分辨率等级;在低分辨率尺度空间中,用灰度类间最大方差和类内最小方差共同构造适应度函数,采用改进的遗传算法对各帧道路图像进行阈值自适应分割,找到准确的道路边界,最近两边界中心位置即机器人行驶方向。采用小型陆地自主车作为研究平台,并在卡耐基梅隆大学（CMU）提供的室外移动机器人道路视频中进行算法测试。结果 本文方法能够在具有阴影、裂纹、光照度变化的道路条件下鲁棒分割出道路边界,机器人可以平均30 km/h的速度在有较严重阴影干扰的校园道路上行驶,视觉系统的处理速度平均可达到20 ms/帧。结论 本文方法比传统的灰度直方图分割法表现出更强的环境自适应性,可实现较为鲁棒的室外道路检测,并可作为室外自主移动机器人非结构化道路检测的一种鲁棒性较强的方法加以推广。     

基于多核支持向量机集成的智能玻璃制品检测算法

研究提出了新的玻璃制品智能检测系统和算法;根据玻璃制品检测的需要,设计了一个机器视觉检测系统,并开发了实验样机;在获取玻璃制品图像后,根据缺陷的特点来分割出可能缺陷区域,然后在可能缺陷区域内提取缺陷特征;提出采用一种多核函数支持向量机集成方法来对特征进行分类;此多核函数支持向量机集成采用遗传算法来协同优化集成中支持向量机的各项参数,使得各支持向量机在拥有较高分类性能的同时保持差异性;而在最后集成各支持向量机时采用了遗传选择集成方法;实验表明采用文中提出的检测算法在实验样机上检测玻璃制品质量,准确率可达97%以上.     

机器视觉表面缺陷检测综述

目的 工业产品的表面缺陷对产品的美观度、舒适度和使用性能等带来不良影响,所以生产企业对产品的表面缺陷进行检测以便及时发现并加以控制。机器视觉的检测方法可以很大程度上克服人工检测方法的抽检率低、准确性不高、实时性差、效率低、劳动强度大等弊端,在现代工业中得到越来越广泛的研究和应用。方法 以机器视觉表面缺陷检测为研究对象,在广泛调研相关文献和发展成果的基础上,对基于机器视觉在表面缺陷检测领域的应用进行了综述。分析了典型机器视觉表面缺陷检测系统的工作原理和基本结构,阐述了表面缺陷视觉检测的研究现状、现有视觉软件和硬件平台,综述了机器视觉检测所涉及到的图像预处理算法、图像分割算法、图像特征提取及其选择算法、图像识别等相关理论和算法研究,并对每种主要方法的基本思想、特点和存在的局限性进行了总结,对未来可能的发展方向进行展望。结果 机器视觉表面缺陷检测系统中,图像处理和分析算法是重要内容,算法各有优缺点和其适应范围。如何提高算法的准确性、实时性和鲁棒性,一直是研究者们努力的方向。结论 机器视觉是对人类视觉的模拟,机器视觉表面检测涉及众多学科和理论,如何使检测进一步向自动化和智能化方向发展,还需要更深入的研究。     

图像匹配方法研究综述

目的 图像匹配作为计算机视觉的核心任务,是后续高级图像处理的关键,如目标识别、图像拼接、3维重建、视觉定位、场景深度计算等。本文从局部不变特征点、直线、区域匹配3个方面对图像匹配方法予以综述。方法 局部不变特征点匹配在图像匹配领域发展中最早出现,对这类方法中经典的算法本文仅予以简述,对于近年来新出现的方法予以重点介绍,尤其是基于深度学习的匹配方法,包括时间不变特征检测器（TILDE）、Quad-networks、深度卷积特征点描述符（DeepDesc）、基于学习的不变特征变换（LIFT）等。由于外点剔除类方法常用于提高局部不变点特征匹配的准确率,因此也对这类方法予以介绍,包括用于全局运动建模的双边函数（BF）、基于网格的运动统计（GMS）、向量场一致性估计（VFC）等。与局部不变特征点相比,线包含更多场景和对象的结构信息,更适用于具有重复纹理信息的像对匹配中,线匹配的研究需要克服包括端点位置不准确、线段外观不明显、线段碎片等问题,解决这类问题的方法有线带描述符（LBD）、基于上下文和表面的线匹配（CA）、基于点对应的线匹配（LP）、共面线点投影不变量法等,本文从问题解决过程的角度对这类方法予以介绍。区域匹配从区域特征提取与匹配、模板匹配两个角度对这类算法予以介绍,典型的区域特征提取与匹配方法包括最大稳定极值区域（MSER）、基于树的莫尔斯区域（TBMR）,模板匹配包括快速仿射模板匹配（FAsT-Match）、彩色图像的快速仿射模板匹配（CFAST-Match）、具有变形和多样性的相似性度量（DDIS）、遮挡感知模板匹配（OATM）,以及深度学习类的方法MatchNet、L2-Net、PN-Net、DeepCD等。结果 本文从局部不变特征点、直线、区域3个方面对图像匹配方法进行总结对比,包括特征匹配方法中影响因素的比较、基于深度学习类匹配方法的比较等,给出这类方法对应的论文及代码下载地址,并对未来的研究方向予以展望。结论 图像匹配是计算机视觉领域后续高级处理的基础,目前在宽基线匹配、实时匹配方面仍需进一步深入研究。     

Model-based inspection system for component boards

A computer vision system has been developed to perform the complex inspection of many components on a flat surface or board. It has a hierarchical structure which allows it to determine the location of components with one camera resolution and then to inspect for small details with a higher resolution. The system has generality in that the same program can be used to inspect boards with completely different components by forming a new model. The MODEL-FORMER program carries out an interactive session with an operator who specifies the layout of components, showing the system each different component and selecting inspection tests which apply to specific regions of the component. These tests allow the INSPECTOR system to differentiate between good and bad components. During inspection, the system (using gray-scale rather than the more limited binary techniques) automatically inspects all components on the board for: (1) presence of correct component; (2) correct position and orientation; (3) any other important features which the operator has designated.     

基于十字线结构光视觉测量系统

为了将线结构光视觉测量原理应用于零件二维轮廓的同时检测，文中使用了十字线激光器作为结构光投射器，它将在空间中形成相互垂直的两个光平面，配置一个CCD相机就组成了一个十字线结构光视觉测量系统。基于计算机视觉技术，文中重点介绍了该视觉测量系统的标定参数及方法，利用光平面的共面性推导了系统的测量原理，建立了两个光平面相应的数学模型，并用其实现了对零件的非接触二维轮廓曲线的测量，验证了模型的正确性和可行性，且方便快捷。     

Boundary Value Technique for Finding Numerical Solution to Boundary Value Problems for Third Order Singularly Perturbed Ordinary Differential Equations

A class of singularly perturbed two point boundary value problems (BVPs) for third order ordinary differential equations is considered. The BVP is reduced to a weakly coupled system of one first order Ordinary Differential Equation (ODE) with a suitable initial condition and one second order singularly perturbed ODE subject to boundary conditions. In order to solve this system, a computational method is suggested in this paper. This method combines an exponentially fitted finite difference scheme and a classical finite difference scheme. The proposed method is distinguished by the fact that, first we divide the domain of definition of the differential equation into three subintervals called inner and outer regions. Then we solve the boundary value problem over these regions as two point boundary value problems. The terminal boundary conditions of the inner regions are obtained using zero order asymptotic expansion approximation of the solution of the problem. The present method can be extended to system of two equations, of which, one is a first order ODE and the other is a singularly perturbed second order ODE. Examples are presented to illustrate the method.     

Sulla soluzione dell'equazione biarmonica con metodi di programmazione lineare

The solution of biharmonic equation is studied with linear programming methods to obtain an a posteriori estimate of truncation errors. For this the algorithm proposed by Barrodale and Young has been adopted, which adjusts to the peculiar problem's structure the simplex method, and a lowest memory occupation requires, avoiding any constraints duplication. Carrying out some different boundary points distribution, a comparison is made among the results obtained by minimizing the error according to two different norms. It's pointed out that withL ₁ norm, we can reach a simpler and shorter problem formulation, and a solution which offers a better punctual approximation than we could obtainwithL _∞ norm, which supplies at contrary a better approximation. It's important to point out that such algorithm, when we try to approximate withL _∞ norm the biharmonic equation solution generates a cyclic procedure; so the ‘lexicographic method’ variant has been introduced. Numerical tests carried out are exposed in the following tables.     

A new method for implementing moment functions in a CMOS retina

We present in this paper a new method for implementing geometric moment functions in a CMOS retina. The principle is based on the similarity between geometric moment equations and the measurement of the correlation value between an image to analyze and a range of grey levels. The latter is approximated by a binary image called mask using a dithering algorithm in order to reduce hardware implementation cost. The correlation product between the mask and the image under analysis gives an approximated value of the geometric moment with an error less than 1% of the exact value. Finally, the results obtained by our approach have been applied to an object localization application and the localization error due to the approximated moment values reported. Olivier Aubreton was born in Vichy on August 31, 1973. He obtained the agrégation examination in June 2000 and received the D.E.A. degree (equivalent to a master degree) in image processing in June 2001. He is currently a lecturer working towards a Ph. D. degree at Laboratory LE2I in the IUT of Le Creusot in Burgundy. His research interests include the design, development implementation, and testing of silicon retinas for pattern matching and pattern recognition. Lew F.C. Lew Yan Voon received his Ph.D. degree in Computer Aided Design of VLSI circuits from Montpellier University, France, in March 1992. Since September 1993, he has been first assistant professor and then associate professor at the University of Burgundy. His research interests lie in the field of pattern recognition and in the design of silicon retinas in standard CMOS technology for real-time inspection by machine vision. Bernard Lamalle was born in Autun on May 1, 1946. He obtained the Ph.D. degree in 1973 from the Université de Bourgogne in Dijon. During 1980 to 2000 he has been Maître de conférences at the IUT of Le Creusot in Bourgogne. He joined the image processing team of laboratory Le2i in 1992. Since 2000, he has been appointed full professor of the University of Bourgogne. His field of interest is principally the study and design of silicon retinas dedicated to industrial control. He has in charge some industrial contracts in the field of quality control by artificial vision and he holds two patents in the field of image processing and smart sensors. Guy Cathébras was born in Uzès, France, in 1961. He received the French engineer degree from the Ecole Nationale Supérieure de l'Electronique et de ses Applications, Cergy, France, in 1984 and the Diplôme de Doctorat de l'Université de Montpellier, France, in 1990. Since 1992 he is an assistant professor of microelectronics at the Institut des Sciences de l'Ingénieur de Montpellier. His current research interests include the design of imagers and silicon retinas using standard CMOS technologies.