Particle reconstruction of volumetric particle image velocimetry with strategy of machine learning
This addresses particle reconstruction in volumetric particle image velocimetry (a fluid dynamics measurement technique), offering practical enhancements for researchers in experimental fluid mechanics.
The paper tackles the under-determined inverse problem of 3D particle reconstruction from limited 2D projections by proposing a convolutional neural network (CNN) with geometry-informed features to refine coarse initial guesses from traditional methods. The result shows significant improvements in reconstruction quality, robustness to noise, and at least an order of magnitude faster offline processing compared to existing algebraic reconstruction technique (ART)-based algorithms.
Three-dimensional particle reconstruction with limited two-dimensional projections is an under-determined inverse problem that the exact solution is often difficult to be obtained. In general, approximate solutions can be obtained by iterative optimization methods. In the current work, a practical particle reconstruction method based on a convolutional neural network (CNN) with geometry-informed features is proposed. The proposed technique can refine the particle reconstruction from a very coarse initial guess of particle distribution generated by any traditional algebraic reconstruction technique (ART) based methods. Compared with available ART-based algorithms, the novel technique makes significant improvements in terms of reconstruction quality, {robustness to noises}, and at least an order of magnitude faster in the offline stage.