Laser Beam Optimization for TDLAS Tomography From Asymptotic Point Spread Functions

Tunable diode laser absorption spectroscopy (TDLAS) tomography is widely used for gas temperature and concentration imaging. Laser beam optimization is an efficient way to improve imaging quality with limited available projections for tomography. A beam optimization method independent of any specific distribution was proposed. Asymptotic point spread functions (PSFs) were generated with the Dirac function approximated by Gaussian functions, and the shape proximity between PSF and the Dirac function was estimated. Simultaneous algebraic reconstruction technique (SART) was applied for fan-beam layout optimization. Typical beam optimization methods were compared for numerical verifications. Reconstruction errors of the optimal beam layouts obtained from the proposed method were generally smaller than or equivalent to those from the literature. A TDLAS tomographic sensor with the optimal layout was realized, with only five pairs of distributed feedback (DFB) lasers used for nine projection angles. The flame of a Bunsen burner with a flat nozzle was detected, and beam layouts from the real sensor with projections missing were evaluated for experimental verifications. Averaged deteriorations of image quality indicated by the structural similarity index (SSIM) of temperature and H2O mole fraction were only 0.7% and 1.5%, implying the immunity of missing projections by the optimal beam layout. The SSIM variation with the number of missing projections was also well predicted by the proposed method, which verified the correlation between shape proximity and image quality.