Experimentally calibrated weight-matrix-coupled GFP-SART algorithm for 3D particle field reconstruction through light field imaging

Abstract

Accurate three-dimensional (3D) flow field measurement through light field particle image velocimetry (LF-PIV) relies on the precise reconstruction of tracer particle positions and luminescent intensities. This accuracy depends heavily on the performance of the tomographic reconstruction algorithm and the fidelity of the associated weight matrix. This paper proposes a novel method—ECWM-GFP-SART—that enhances the reconstruction accuracy by integrating 3D Gaussian fitting positioning (GFP) into the simultaneous algebraic reconstruction technique (SART) algorithm to mitigate axial elongation of particles, and by constructing an experimentally calibrated weight matrix (ECWM) to correct deviations between theoretical and actual imaging models. The GFP method improves particle positioning by applying Gaussian fitting to the intensity distribution of elongated particles, effectively eliminating the elongation artifacts. Meanwhile, the ECWM is established by analyzing the ratio between the grayscale values of light field images obtained from the point light sources to their actual luminescent intensities, producing an accurate weight matrix consistent with experimental conditions. The effectiveness of the proposed ECWM-GFP-SART method is evaluated by reconstructing 3D particle fields with concentrations ranging from 0.20 to 0.80 particles per micro-lens (ppm), and compared against the traditional SART algorithm. Results demonstrate that ECWM-GFP-SART significantly improves the weight matrix accuracy and effectively eliminate the axial elongation of particles. Compared to the traditional SART, ECWM-GFP-SART increases the reconstruction quality coefficient from 0.10 to over 0.92 and accelerates the reconstruction speed by a factor of four.