Radial Basis Function Coupled SART Method for Dynamic LAS Tomography

Abstract

Laser absorption spectroscopy (LAS) tomography is an effective combustion diagnosis method for its simultaneous imaging of temperature and gas molar concentration. Only limited projections are available in many applications, the unknowns in the image reconstruction are much more, and it restricts the quality of reconstructed images. The radial basis function (RBF) is introduced to simplify the expressions of continuous distributions in the region of interest (ROI). The number of unknowns is greatly reduced, as gray values in the imaging region are depicted by the coefficient vector of these RBFs. The continuity in temperature distributions was coupled with the simultaneous algebraic reconstruction technique (SART) method. The coefficient vector of these RBFs and density distributions of integral absorbances were obtained. Gaussian compactly supported RBF was selected to reduce computation burden and reconstruction error. Performance comparisons between the proposed method and the classical SART method were made. The imaging quality of the proposed method was superior to SART, the temperature error was smaller than SART by nearly 10% and high spatial resolution was achieved. Dynamic flames in the McKenna burner and wind tunnel were used to verify its robustness and applicability in harsh cases.