基于超限化的吸收层析反演算法研究

针对吸收光谱层析（tomographic absorption spectroscopy，TAS）中的反问题，利用超限化（superiorization）在重建中引入平滑性、稀疏性等先验条件，对现有的层析反演算法提出了改进.通过计算机仿真，对代数重建（algebraic reconstruction technique，ART）算法、最大似然期望最大化（maximum likelihod expectation maximization，MLEM）算法的超限化在TAS反演中的应用进行了研究.在仿真中，对比了不同二维火焰场、不同目标约束函数下超限化算法的效果，研究了噪声对重建的影响以及超限化算法在不同条件下的计算效率.研究结果证实了超限化对于层析反演算法在计算精度、收敛速度等方面的提升效果. 

On the Superiorization of Inversion Algorithms for Tomographic Absorption Spectroscopy

Superiorization,which introduces a priori conditions such as smoothness and sparseness to flame reconstruction,was applied to tomographic absorption spectroscopy(TAS)in this paper,as an improvement of existing tomographic inversion algorithms.Through simulation of TAS,the superiorization of algebraic reconstruction technique(ART)and maximum likelihood expectation maximization(MLEM)were studied.The performances of superiorized algorithms for different two-dimensional flame fields and under different target constraints were compared.The influence of noise on the reconstruction,as well as the computational efficiency of the superiorized algorithms under different conditions was studied.The results suggest that the superiorization can improve the tomographic inversion algorithms in terms of reconstruction accuracy and convergence rate.