在工业生产中大型工件的三维测量问题上，由于传统测量设备测量范围的限制，常常需要一种三维数字化处理方法来快速得到其局部几何数据，再通过曲面拟合及拼接完成实际测量。本文研究开发的标记目标的视觉测量系统可用于实现三维数据测量。 文中首先对三维视觉测量技术及现有测量系统进行分析研究，完成了标记目标的视觉测量系统分析与设计。系统功能主要有：参数预置、图像预处理、目标中心定位、特征点匹配、相机定标、三维重建和异常点剔除。其中重点对目标中心定位、特征点匹配、相机定标及三维重建等关键模块涉及的算法进行深入研究，通过分析比较目前流行设计方法的优缺点，从而确定了本系统关键算法的选择。 在系统实现中选择Microsoft Visual C++ 6.0作为开发平台，提高了用户交互操作的易便性。但是由于目标中心定位、特征点匹配、相机定标和三维重建等关键模块涉及大量图像处理和数学矩阵运算，所以这几个模块由Matlab开发完成，降低了系统开发难度并缩短了开发周期。在关键模块接口的实现上，文中首先分析比较了VC++与Matlab常用混合编程方法，然后重点通过研究基于COM组件的混合编程说明了本系统关键模块接口的实现过程，进而完成了标记目标的视觉测量系统开发。 最后通过汽车模具检测和雕塑检测两个应用实验及其结果分析给出了本文研究结论，证实本系统能够减少测量过程中对硬件设备、测量环境和专业技术的依赖，而且能够得到稳定性高、精度良好的三维测量结果，可应用于监测点形变计算、产品质量检测及逆向工程等多个领域。

In industrial production of large-size workpiece three-dimensional measurement issues, because of the limitation of traditional measuring equipment measuring range, The actual measurement often requires a three-dimensional digital method to get the local geometric data quickly at first, and then completed by surface fitting and stitching. In this paper, the development of marked-targets visual measurement system can be used to implement three-dimensional data measurement. First to do is to analyses three-dimensional visual measurement technology and the existing measurement system, and complete the analysis and design of marked -targets visual measurement system. The major functions contain parameters presetting, image preprocessing, target center location, feature matching, camera calibration, three-dimensional reconstruction and outliers rejection. The next focus on the key modules of the target center location, feature matching, camera calibration and three-dimensional reconstruction and involved algorithms in-depth study. And determine the key algorithm of the system by analyzing and comparing the advantages and disadvantages of the current popular design method. The system uses Microsoft Visual C++ 6.0 as the development platform to improve the user interaction. However, due to the key modules of the target center location, feature matching, camera calibration, and three-dimensional reconstruction involving a large number of image processing and mathematical matrix operations, the modules were developed with the aid of Matlab. This reduces the difficulty of system development and shortens the development cycle. On the implementation of the key modules interface, the paper begin with an analysis and compares of the VC++ and Matlab mixed programming method, and then focus on introducing the implementation process of the system key modules interface by mixed processing based on COM component, then complete the marked-targets visual measurement system. Finally, the paper gives the conclusion of this study by two sets of images of the automobile molds and sculptures and the result. Experience shows the system can reduce the dependence on hardware devices, measurement environment and professional technology, and achieve high stability, high accuracy three-dimensional measurement results, it can applied to deformation calculation of the monitoring points, product quality inspection and reverse engineering and other fields.