Ha Minhtuan, Xiao Changyan, Pham Dieuthuy, Ge Junhui
Appl Opt. 2020 Mar 20;59(9):2674-2685. doi: 10.1364/AO.381149.
Structured light 3D reconstruction methods using a De Bruijn sequence-based color grid pattern have an impressive advantage of fast and accurate decoding, which leads to fast 3D reconstruction. They are especially suitable for capturing moving objects. However, the drawback of these methods is their high false decoding rate while dealing with feature points at the object's boundaries, and objects can be prone to becoming deformed by the uneven structure of the dynamic scene. To solve this problem, we present an efficient opened-grid-point detector and a complete grid pattern decoding method. Specifically, a new, to the best of our knowledge, color grid pattern is designed to reduce the influence of color noise and increase the density of 3D cloud points. In addition, a LCD screen projected with the proposed pattern is utilized to calibrate the camera-projector system. The experiments, conducted in a laboratory without a light curtain, demonstrate that the proposed method can fully satisfy the requirements of real applications.
使用基于德布鲁因序列的彩色网格图案的结构光三维重建方法具有快速准确解码的显著优势，这使得三维重建速度很快。它们特别适合捕捉移动物体。然而，这些方法的缺点是在处理物体边界处的特征点时误解码率很高，并且物体容易因动态场景的不均匀结构而变形。为了解决这个问题，我们提出了一种高效的开放网格点检测器和一种完整的网格图案解码方法。具体来说，据我们所知，设计了一种新的彩色网格图案，以减少颜色噪声的影响并增加三维云点的密度。此外，使用投射有该图案的液晶显示屏来校准相机-投影仪系统。在没有光幕的实验室中进行的实验表明，所提出的方法能够充分满足实际应用的要求。
