连接件平整度检测是保障其安装质量的重要手段，具有“三维信息”、“复杂反射率”的特征以及“高精度”、“人机交互友好”的要求。但目前市场上平整度仍缺乏快速有效的测量方法，利用传统方法实现高效检测存在一定困难，相比之下，结构光技术具有高速、高精度、高点云密度、非接触等优点，能够很好地适用于平整度检测。因此本文意在利用结构光三维测量技术，实现连接件平整度高精度检测；同时，考虑到测量范围内存在大量连接件的可能，为提高不同连接件的辨识度，本文拟通过增强现实实现连接件与平整度的一一对应，避免人工产生误判，提高使用体验。 高精度三维重建是连接件平整度高精度检测的基础，针对此要求，本文分析了结构光三维测量系统的精度影响因素：标定、编码以及硬件，分别选用传统标定方法与时间相移编码方法以保证结构光系统计算与解码精度；并特别设计系统结构，利用投影仪偏心投影的特性，将相机放置在投影仪虚拟光心位置最大化两者公共视野，充分利用相机分辨率以提高系统测量精度。最终本文搭建了一套结构光三维测量系统，能够实现目标的高精度三维重建，为后续平整度分析打下基础。 由于连接件及其邻域材料存在多样性，表面反射率较为复杂，容易导致结构光测量过程发生过曝而无法正常三维重建。本文提出了结构光动态投影理论解决易过曝物体测量问题，利用相机采集图像的反馈信息动态修改投影图案，降低过曝区域的投影强度，实现过曝物体的三维重建。实验证明结构光动态投影理论能够有效降低易过曝物体测量过程中的过曝点数量，实现了复杂反射率对象的完整三维重建。 最后，为实现连接件平整度完整检测流程并提高辨识度，本文利用上述结构光三维测量系统，对连接件平整度进行检测并在对应位置应用增强现实技术投影显示。首先通过结构光技术获得目标三维点云；然后结合连接件特点，利用图像处理与特征提取等技术对连接件及其邻域进行点云分割；接着计算连接件平整度；最后基于已有的结构光硬件设备，生成平整度虚拟模型，投影至连接件真实世界对应位置，实现增强现实显示。实验证明本文提出的结构光连接件平整度在线检测方法具有较高的精度与稳定性，同时其投影显示效果能够辅助人们对不同连接件平整度测量结果进行区分，实现了连接件平整度的在线检测与增强现实显示。

The flatness detection of the connecting piece is an important means to ensure the quality of the installation. It has the characteristics and requirements of“high precision”,“complex reflectivity”,“3D information”and“friendly human-computer interaction”. However, fast and effective measurement methods for flatness detection are still rare in the market. It is difficult to achieve high-efficiency detection by traditional methods. In contrast, structured light technology has the advantages of high speed, high precision, high point cloud density, non-contact, which is beneficial for flatness detection. Therefore, this paper intends to use the structured light-based 3D measurement technology to achieve high-precision detection of the the connector flatness within a certain range. At the same time, considering the possibility of a large number of connectors in the range, in order to improve the identification of different connectors, this paper intends to achieve one-to-one correspondence between every connector and its flatness through augmented reality display, so as to avoid artificial misjudgment and improve the user experience. High-precision 3D reconstruction is the basis for the high-precision detection of theconnector flatness. In view of this requirement, this paper analyzes the factors affecting the precision of the 3D measurement system of structured light: calibration, coding and hardware. The traditional calibration method and temporal phase-shifting method are selected respectively to ensure the calculation and decoding accuracy of the system. By utilizing the characteristics of the eccentricity of projector, a camera is specially placed at the virtual optical center of the projector to maximize the public field of view. Hence the camera resolution is fully utilized to improve the measurement accuracy of the system. Finally, a 3D measurement system based on structured light is built? which can realize the high-precision 3D reconstruction of the target and lay a foundation for the surface quality detection. Due to the diversity of connector and adjacent materials, the surface reflectivity iscomplex, which easily leads to the over-exposure of structured light measurement process and the failure of 3D reconstruction. In this paper, dynamic projection theory of structural light is proposed to solve the problem of measuring over-exposed objects, and the feedback information of images collected by the camera is used to modify the projection pattern dynamically, so as to reduce the projection intensity of over-exposed areas and realize the 3D reconstruction of over-exposed objects. Experimental results show that the dynamic projection theory of structured light can effectively reduce the number of over-exposed points in the measurement process of easily over-exposed objects and realize the complete 3D reconstruction of industrial products with complex reflectivity. Finally, in order to realize the complete detection process of the connector flatness and improve the recognition degree, this paper uses structured light-based 3D measuring system to detect the connector flatness such as screws and display the detected results at the corresponding position. Firstly, the target 3D point cloud is obtained by structuredlight technology. Combining the characteristics of the connecting parts, the point cloud is separated by the image processing and feature extraction techniques, and then the connector flatness of is calculated. Finally, a flatness virtual model is generated, and based on the structured light system, the hardware is used to project the detected results to the corresponding position of the connector in real world to realize the augmented reality display. The experiment proves that the online connector flatness detection method by structured light system in this paper has high accuracy and stability. At the same time, there is a good display effect of augmented reality, which can help people distinguish the measurement results of different connector flatness. Finally, both the on-line connector flatness detection and augmented reality display are realized.