Neural refractive index field: Unlocking the Potential of Background-oriented Schlieren Tomography in Volumetric Flow Visualization

Yuanzhe He,Yutao Zheng,Shijie Xu,Chang Liu,Di Peng,Yingzheng Liu,Weiwei Cai

2024-09-23

Abstract:Background-oriented Schlieren tomography (BOST) is a prevalent method for visualizing intricate turbulent flows, valued for its ease of implementation and capacity to capture three-dimensional distributions of a multitude of flow parameters. However, the voxel-based meshing scheme leads to significant challenges, such as inadequate spatial resolution, substantial discretization errors, poor noise immunity, and excessive computational costs. This work presents an innovative reconstruction approach termed neural refractive index field (NeRIF) which implicitly represents the flow field with a neural network, which is trained with tailored strategies. Both numerical simulations and experimental demonstrations on turbulent Bunsen flames suggest that our approach can significantly improve the reconstruction accuracy and spatial resolution while concurrently reducing computational expenses. Although showcased in the context of background-oriented schlieren tomography here, the key idea embedded in the NeRIF can be readily adapted to various other tomographic modalities including tomographic absorption spectroscopy and tomographic particle imaging velocimetry, broadening its potential impact across different domains of flow visualization and analysis.

Fluid Dynamics,Human-Computer Interaction,Machine Learning

What problem does this paper attempt to address?

This paper attempts to solve the problems encountered in Background - Oriented Schlieren Tomography (BOST) technology in three - dimensional flow field visualization. Specifically, the traditional voxel - based meshing scheme has the following challenges:

1. **Insufficient spatial resolution**: Due to the limited number of voxels, a higher spatial resolution cannot be achieved.
2. **Significant discretization error**: The voxelization method assumes that the physical properties within each voxel are constant, which will lead to severe reconstruction artifacts.
3. **Poor noise immunity**: The voxelization method is sensitive to noise and is prone to introducing errors.
4. **Excessively high computational cost**: As the number of voxels increases, the weight matrix and memory requirements grow exponentially, resulting in huge consumption of computational resources.

To solve these problems, the paper proposes an innovative reconstruction method - Neural Refractive Index Field (NeRIF). This method implicitly represents the flow field by training a lightweight neural network, receiving the spatial coordinates of points within the reconstruction domain as input, and outputting the corresponding refractive index and its gradient. The main advantages of NeRIF include:

1. **High spatial resolution**: Representing the flow field through a neural network avoids the discretization of voxelization, making the spatial resolution close to the camera pixel resolution.
2. **Reduction of discretization error**: A random sampling strategy is adopted to approximate continuous integration and differentiation, effectively reducing the discretization error.
3. **Reduction of memory requirements**: There is no need for a huge coefficient matrix, which significantly reduces the memory requirements.
4. **Prevention of over - fitting**: The carefully designed loss function minimizes the inherent mathematical differences between the output refractive index and its gradient, effectively preventing over - fitting to measurement noise.

The paper demonstrates the application effect of NeRIF in turbulent Bunsen flames through numerical simulation and experimental verification. The results show that NeRIF can significantly improve the reconstruction accuracy and spatial resolution while reducing the computational cost. In addition, the core idea of NeRIF can also be applied to other tomographic modalities, such as tomographic absorption spectroscopy and tomographic particle image velocimetry, and has broad application prospects.