Neural refractive index primitives for flame field reconstruction using background-oriented schlieren

An improved neural implicit refractive-index reconstruction framework for background-oriented schlieren tomography is presented, enabling accurate and efficient three-dimensional recovery of refractive-index fields using a compact neural representation. The method adopts the refractive-index field as the sole neural primitive and integrates encoding strategies, field gradient losses, a three-dimensional mask constraint, and a two-dimensional mask ray-sampling scheme to achieve fast convergence and spatially coherent reconstructions. By jointly enforcing these components, the proposed framework maintains a simple model structure while improving optimization stability and reconstruction accuracy. Comprehensive evaluations on synthetic and experimental flame datasets across varying turbulence intensities demonstrate robust performance and accurate recovery of refractive-index distributions, as well as temperature-related structures under restricted assumptions.