Iterative Reconstruction Method via Adaptive Grid Partition for TDLAS Tomography of Reactive Flows

Abstract Limited projections from actual optical paths yield an underdetermined projection equation for tunable diode laser absorption spectroscopy (TDLAS) tomography. An iterative reconstruction method via adaptive grid partition was proposed to solve the equation. A partition of grids was used to determine equality constraints on grids for the same pixel value, and the unknows were decreased. Projection equation of a gridding in high spatial resolution was decomposed to several equations of less unknows. It was verified by using asymptotic point spread functions. The shape proximity of the point spread function to Dirac function with the proposed method was 21.1% smaller than that of simultaneous algebraic reconstruction technique (SART). Also, smaller image errors and higher structure similarity index (SSIM) values were achieved in both noise-free and noisy cases compared to SART. In temperature imaging on a high-temperature wind tunnel, a relative deviation of 2.55% from thermocouple values verified the accuracy of the proposed method.