基于改进的格雷码法的三维重建系统

A Three-Dimensional Reconstruction System Based on Improved Gray Code Method

摘要: 随着结构光三维测量技术的迅速发展，对精度的要求日益提高。基于格雷码的方法具有鲁棒性强和抗噪性良好的优点，易于实现，在工业检测、生物医学等领域被广泛应用。但传统的基于格雷码图案的相位展开方法往往容易出现跳变误差，为了避免这一情况，提高相位展开的精度，设计了一种六步相移结合互补格雷码的方法。首先对测量系统进行标定，然后将设计好的光栅图投影到待测物体上，经过处理得到连续相位图，结合标定数据完成物体的三维重建。增加的互补格雷码与传统格雷码存在半个周期的错位，可以有效避免边界解码发生的错误。相比较传统的投影一黑一白的图案生成掩码图像，本文采取最大类间方差法对全白图案进行处理得到掩码图像，在减少投影图片的同时，能够很好地提取感兴趣区域，提高相位展开的速度。利用标准多圆柱进行实验，结果表明：系统的重建精度在0.3 mm以内，并且对更为复杂的大卫头像模型进行实验，较好地重建出了待测物体。

Abstract: With the rapid development of structured light 3D measurement technology, the demand for ac-curacy is increasing. The method based on Gray code has the advantages of strong robustness and good anti-noise performance, and is widely used in industrial inspection, biomedical and other fields for its ease of implementation. However, the traditional phase unwrapping method based on Gray code patterns often results in jump errors. To avoid this situation and improve the accuracy of phase unwrapping, a six-step phase shift method combined with complementary gray code is designed. First, the measurement system is calibrated, and then the designed grating pattern is projected onto the object to be measured to obtain a continuous phase map. Combined with the calibration data, the 3D reconstruction of the object is completed. The added complementary Gray code is offset from the traditional Gray code, which can effectively avoid decoding errors at the boundary. Compared with the traditional method of generating masking images with one black and one white pattern, this article processes the full white pattern using the maximum inter-class variance method to obtain a masking image, which can reduce the projection images and extract the regions of interest well, improving the speed of phase unwrapping. The experimental results using standard multi-cylinder show that the reconstruction accuracy of the system is within 0.3 mm, and the more complex gypsum-like objects is well reconstructed.