Abstract—Three-dimensional coordinate measurement of fea-ture points on the surface of a large-scale workpiece is impor-tant and difficult. Various relative measuring methods have beenpresented in recent years, and the machine vision method hasbeen paid more attentions by researchers. The application of themachine vision method in 3-D coordinate measurement of featurepoints on the surface of a large-scale workpiece is discussed inthis paper, and an accurate, simple, and new measuring methodis proposed. The design of the measuring system mainly considersthe following aspects: 1) the principle and composition of themeasuring system; 2) the monocular vision algorithm for cameralocating; 3) the calibration algorithm of the charge-coupled device(CCD) camera; 4) the image processing algorithm of cross-cuttingfeature points and the calculation of their 2-D image coordinates;and 5) the binocular stereo vision algorithm for depth measure-ment based on the large-scale coordinate measuring machine.The experimental results indicate the correctness and reliabilityof the new measuring method, and we believe that it will be areliable and efficient technique for the noncontact 3-D coordinatemeasurement of cross-cutting feature points on the surface of alarge-scale workpiece.Index Terms—Camera calibration, coordinate measuringmachine (CMM), feature point, monocular vision, stereo vision,3-D coordinate measurement.I. 34623
INTRODUCTIONIN THE manufacturing process of some kinds of large-scale workpieces in mechanical industry, 3-D coordinateprecision measurement of feature points on the surface of alarge-scale workpiece is always needed to evaluate the contourquality, overall assembling quality, etc. The relative measureddimensions of a large-scale workpiece can also be calculated,and more important, the measuring results can be fed back tothe manufacturing process in time. For example, sometimes,several cylindrical segments are mounted together to form aManuscript received September 1, 2008; revised March 15, 2009; acceptedMarch 16, 2009. Date of publication October 13, 2009; date of current versionJune 9, 2010.S. Zhu is with the Department of Measurement Control and InformationTechnology, School of Instrumentation Science and Optoelectronics Engi-neering, Beihang University, Beijing 100191, China (e-mail: shiping.zhu@buaa.edu.cn).Y. Gao was with the Département de Génie Électrique et de Génie Informa-tique, Faculté de Génie, Université de Sherbrooke, Sherbrooke, QC J1K 2R1,Canada. She is now with Beijing Hollysys Co., Ltd., Beijing 100096, China(e-mail: gaoyang@hollysys.com).Color versions of one or more of the figures in this paper are available onlineat http://ieeexplore.ieee.org.Digital Object Identifier 10.1109/TIM.2009.2030875long cylindrical workpiece, and the concentricity of all thesecylindrical segments needs to be measured to evaluate theassembly quality. To realize the 3-D coordinate precision mea-surement and further the measurement of concentricity, thefeature points on the surface of a workpiece are usually needed.Generally, two types of feature points can be used in theactual manufacturing process: 1) the noncutting feature point(such as projecting light spot, light fringe [1], [2], light grid[3], and light pattern [4]) and 2) the cutting feature point at themeasured position on the surface of a workpiece. No matterwhich type of feature point is used, the measurement precisionof concentricity depends on the measurement precision of the3-D coordinates of feature points.In the proposed measuring system, the intersection point ofcross-cutting lines is used as the feature point. The cross-cuttinglines are traced along the transversal and longitudinal directionsand are cutting on the surface of a large-scale workpiece. Notethat they are not located along the junctions of each cylindricalsegment, but they are only located on the surface of each cylin-drical segment according to the measuring requirements.
Theintersection point of the transversal and longitudinal directioncutting lines is recognized as the measured feature point. Animage of the cross-cutting lines and their intersection featurepoint is shown in Fig. 6.Because the dimension of the cylindrical workpiece is rela-tively large, the quantity of feature points is also large, and theyare often quite discretely distributed. All these bring difficultiesfor the 3-D coordinate precision measurement of feature pointson the surface of each cylindrical segment, as well as for theconcentricity measurement of the long cylindrical workpiece.Recently, many 3-D coordinate measuring methods of fea-ture points on the surface of a workpiece have been reported,of which contact and noncontact measuring methods are tworepresentative types. The contact methods mainly contain theplatform feeler gauge, micrometer gauge, inductive trans-ducer, electronic leveling gauge, coordinate measuring machine(CMM), etc. The noncontact methods mainly contain the in-terferometric metrology, autocollimator, electronic theodolite,machine vision technology [5], [6], etc.The platform feeler gauge has widely been used because ofits simplicity, but its measurement precision and efficiency areseriously affected by human factors. The micrometer gauge andinductive transducer are only suitable for a small-area surface. The electronic leveling gauge can only be applied to large-areaplane flatness measurement. The CMM has apparent advan-tages of high precision and high efficiency, but its measuringprobe cannot accurately aim at the intersection feature point ofcross-cutting lines [7], [8].The interferometric metrology and autocollimator have highmeasurement precision, but they need a high-precision large-scale guide rail as the measurement datum. The electronictheodolite is suitable for large-scale space 3-D coordinate mea-surement, but it also cannot accurately aim at the intersectionfeature point of cross-cutting lines. Meanwhile, the problemsof measurement blind areas and the requirement of optimizingthe measuring layout cannot be ignored. Other methods, suchas radar ranging, moire fringe ranging, and tracking-type inter-ferometer ranging, are not suitable for the 3-D coordinate mea-surement of the intersection feature point of cross-cutting lines.The industrial machine vision technology adopts charge-coupled device (CCD) cameras, and it combines the tech-nologies of image processing and precision measurement toperform noncontact 2-D or 3-D coordinate measurement [9].It is applicable to both short and long measuring ranges and hasthe advantages of high precision, high efficiency, high common-ality, low manufacturing cost, etc. It uses a CCD camera as theimage sensor to acquire and recognize the image of each featurepoint; thus, the difficulty of contact-type methods that cannotaccurately aim at the feature point can ultimately be avoided[10]–[12]. The measuring blind areas can be eliminated by theproper arrangement of the geometrical and spatial positions ofCCD cameras.On the basis of the aforementioned discussion, the industrialmachine vision technology and CMM are synthesized in theproposed measuring system, and a new kind of 3-D coordinatemeasuring system of cross-cutting feature points on the surfaceof a large-scale workpiece is presented.II. MEASURING PRINCIPLE AND SYSTEMThe characteristics and specific techniques of the measuringsystem based on the machine vision method and large-scaleCMM and the innovative aspects of the measuring systemwith respect to the existing CCD camera-based machine visionmeasuring systems are described in the list that follows. 机器视觉三坐标测量英文文献和中文翻译:http://www.youerw.com/fanyi/lunwen_32194.html