Three-dimensional combustion front diagnostic method based on down-conversion and light field imaging

To address the challenges associated with data acquisition and processing in three-dimensional (3D) flame-front diagnostics, a down-conversion light field imaging method (DCLF) for intermediate product imaging is developed, enabling non-intrusive measurement of the spatial distribution of intermediate combustion radicals such as OH*. Down-conversion is a photophysical process in which a material absorbs a high-energy, short-wavelength photon and emits one or more lower-energy, longer-wavelength photons. This conversion process improves the imaging efficiency and signal-to-noise ratio of the light field camera. Traditional least squares QR decomposition (LSQR) and conjugate gradient least squares (CGLS) methods suffer from poor noise resistance and significant errors. The DCLF method proposed in this work combines the data compression capabilities of light-field compression and noise reduction (LFCNR) with the regularization advantages of L 1 -norm regularization with coupled prior smoothing (L 1 -PS), thereby overcoming the trade-off between accuracy and efficiency. For 10% random noise, the DCLF method achieves RMSE values of 0.025 and 0.017 for axisymmetric and non-axisymmetric reconstructions, respectively, demonstrating superior accuracy compared to the other four methods. Additionally, under the same noisy conditions, it maintains accuracy comparable to L 1 -PS and reduces the reconstruction time to approximately 10% of that required by L 1 -PS. These results confirm that DCLF facilitates efficient single-shot 3D reconstruction of combustion fronts and provides a robust and high-precision diagnostic tool for complex reacting flows. • A single-shot DCLF framework is developed for 3D reconstruction of flame OH* fields. • A light-field radiative transfer model enables tomographic reconstruction from one exposure. • L 1 -PS achieves RMSE = 0.0094 and NRMSE = 0.018, ∼70% lower than LSQR and CGLS. • DCLF preserves L 1 -PS accuracy while reducing reconstruction time to 10%.