基于多径效应的多相机多投影结构光快速三维测量技术研究

Motivated by the low efficiency in time multiplexing and difficulty in separating superposed patterns in both spatial multiplexing and color multiplexing, a novel multipath-based one-shot multi-camera-multi-projector structured light method is presented for measuring entire 3D objects in this project. By simultaneously projecting encoded patterns, all cameras capture the superposed patterns synchronously. Different from the existing approaches, the proposed method does not need to separate the composed patterns, which are viewed as multipath signals in different transmission delays. In this project, the properties of the superposed patterns are analyzed by multipath effect in wireless communication. The relationship between the superposed patterns and the depth information of objects is explored and the high dynamic range imaging is employed to improve image quality so that the depth information can be obtained directly from the superposed patterns without decomposition. The key target of the project is to propose a one-shot solution and develops an efficient and accurate system for measuring entire surface of a static and moving object, which can be used in industrial inspection, 3D modeling, 3D printing, virtue reality, 3D digitization of Cultural Heritage objects and other applications. The research conducted in this project contributes to the field of 3D modeling and reconstruction and provides a possible solution for measurement of fast moving objects.

针对目前多相机多投影测量方法存在分时投影效率低或重叠图像难于分离的问题，本项目提出多投影仪同时投影多相机同时成像而无需分离重叠图像的快速多相机-投影测量方法。本方法基于无线通信领域的多径效应原理，分析多投影结构光系统重叠信号的特征，探讨混叠信号与物体的深度之间的相关性，利用高动态范围成像技术提高成像质量，实现从混叠信号中直接获取物体的三维信息，为物体的快速、准确全景三维测量提供一种高效的方法。在本项目研究结果的基础上，开发一套适合静态及动态物体的准确三维测量系统。该系统可应用于工业检测、三维建模、3D打印、虚拟现实、文物数字化等众多领域，因此，本项目研究具有重要的理论意义和重大实际应用价值。

目前成熟的视觉系统只提供二维信息。如何让机器能够拥有人类的三维视觉感知能力，并在速度和精度上超越人类，是当今智能化的极大需求，比如自动驾驶的定位导航、无人机的自动避障、虚拟现实的对象建模、工业产品的三维测量等，都是高级机器智能技术在三维视觉上的具体实现和应用。.基于视觉的三维重建是指通过相机获取场景物体的二维图像，并对此图像进行分析处理，再结合计算机视觉理论，利用三维测量技术推导出现实环境中物体的三维信息。在现有三维成像方法中，基于结构光的测量方法是目前应用最广的方法。本项目针对目前结构光技术分时投影效率低或重叠图像难于分离，造成三维重建困难的问题，对三维重建技术的整个环节进行了深入研究，内容包括：高动态范围成像技术研究、图像滤波和增强研究、相机标定方法研究、非线性gamma矫正研究、结构光系统标定研究、编码结构光相位展开研究、三维点云分割、识别及姿态估计研究、相位测量偏折术研究等。在本项目研究结果的基础上，开发了两套快速、准确的结构光三维测量系统，以期用于工业检测、三维建模、3D打印、虚拟现实、文物数字化等众多领域。