Three-dimensional reconstruction of bubble flow field based on multi-camera refraction calibration and improved ordered subset expectation maximization algorithm

Abstract The measurement of the bubble flow field in three-dimensional (3D) space is of great significance for optimizing industrial processes and advancing the development of multiphase flow theory. A calibration method based on the multi-camera refraction calibration (MCRC) method and a 3D reconstruction algorithm of adaptive initialization and constrained projection-based ordered subset expectation maximization (AICP-OSEM) is proposed to reconstruct the bubble flow field. The MCRC method is developed to correct the refraction errors of underwater targets based on the differences between the calibration board images captured in air and in water, avoiding the measurement of the refractive index. The AICP-OSEM algorithm significantly reduces the complexity and time cost of 3D reconstruction by adaptively initializing the voxel value and performing constrained projection during the iteration process. The experimental results demonstrate that both the MCRC method and the AICP-OSEM algorithm can effectively improve the accuracy of the 3D reconstruction results. Furthermore, the AICP-OSEM algorithm can also promote the reconstruction speed of the 3D bubble flow field.