基于五步相移与菱形解相的三维重建方法

A Three-Dimensional Reconstruction Method Based on Five-Step Phase Shift and Rhombus Phase Unwrapping

针对物体三维重建时由于欠采样、噪声、局部阴影、干涉条纹断裂等导致包裹相位变换过快、解包裹相位计算梯度产生不连续和偏差等问题,提出一种采用五步相移法与改进的菱形算法获取物体表面相位信息的方法。结合相位法系统模型,直接根据相位信息计算出物体相机坐标系坐标,获取点云,实现三维重建。实验结果表明,相对于四步相移法以及传统相位解包裹算法,本方法解出的相位展开完全且不失真,特征点提取全面,重建点云表面细节丰富,相位求解精度显著提高。

In order to solve the problems of too fast wrapping phase transformation, discontinuity and deviation of unwrapping phase calculation gradient caused by under-sampling, noise, local shadow and broken interference fringes, a method of obtaining surface phase information of objects by using five-step phase shifting method and improved rhombus algorithm is proposed. Combined with the phase method system model, the camera coordinate system of the object is calculated directly according to the phase information, and the point cloud is obtained to realize three-dimensional reconstruction. The experimental results show that compared with the four-step phase shifting method and the traditional phase unwrapping algorithm, the phase unwrapping solved by the method is complete and undistorted, the feature points are extracted comprehensively, the surface details of the reconstructed point cloud are rich, and the accuracy of phase solution is significantly improved.