Structured-light (SL) three-dimensional (3D) shape measurements technique has been extensively studied and applied in various fields, but the real-time capability of SL system is limited by the number of projected SL patterns. This paper presents a digital image correlation (DIC) assisted phase matching and triple-scanning method for high-speed and accurate 3D shape measurements based on the combined projection of three-step phase-shifting patterns and one speckle pattern. Benefiting from the triple-view geometric constraints and DIC technique, stereo correspondence and absolute fringe order of each valid pixel in the left wrapped phase map can be explored without requiring spatial phase unwrapping process or any priori information, the initial disparity map and absolute phase map can be generated. Subsequently, an efficient disparity refinement scheme based on connectivity area segmentation and neighbor disparity check is applied to the initial disparity map and absolute phase map to compensate the pixels with wrong disparities. Several efficient strategies are developed in the pipeline of data processing to speed up the algorithm. To improve the measurement accuracy, stereo SL model is proposed to make adequate use of triple-view information to calculate 3D coordinates using the disparity map and absolute phase map, and the measurement error has been reduced by 33% compared with the conventional methods. The experimental results demonstrate that the proposed method can perform high-speed and accurate 3D shape measurements for arbitrary surfaces and dynamic scenes.

结构光 (SL) 三维 (3D) 形状测量技术已被广泛研究并应用于各个领域，但 SL 系统的实时能力受到投影 SL 图案数量的限制。本文提出了一种基于三步相移图案和一个散斑图案的组合投影的数字图像相关 (DIC) 辅助相位匹配和三重扫描方法，用于高速准确的 3D 形状测量。得益于三视图几何约束和 DIC 技术，无需空间相位展开过程或任何先验信息、初始视差图和绝对相位图，就可以探索左包裹相位图中每个有效像素的立体对应和绝对条纹顺序可以生成。随后，基于连通性区域分割和邻居视差检查的有效视差细化方案应用于初始视差图和绝对相位图，以补偿具有错误视差的像素。在数据处理管道中开发了几种有效的策略来加速算法。为了提高测量精度，提出了立体SL模型，充分利用三视图信息，利用视差图和绝对相位图计算3D坐标，与传统方法相比，测量误差降低了33%。实验结果表明，所提出的方法可以对任意表面和动态场景进行高速、准确的3D形状测量。