Abstract: To meet the reconstruction requirements of temperature and H₂O concentration distribution in high-temperature flame fields, water vapor was selected as the target measurement gas to reconstruct the temperature and H₂O concentration in combustion flames. Two H₂O absorption lines with wave numbers of 7 153.748 and 7 154.354 cm^-1 were selected. Combined with algebraic reconstruction technique (ART) and adaptive algebraic reconstruction technique (AART) algorithms, simulation reconstruction studies were carried out on the preset multi-peak flame. Results show that the optimal balance state can be achieved when the relaxation factor is set to 0.5 for both algorithms, but the noise resistance performance of the two algorithms is relatively poor. To improve the noise resistance of the algorithms, neighborhood mean smoothing method and Gaussian smoothing method were introduced to optimize AART algorithm. Reconstruction results show that the noise resistance performance of AART algorithm is significantly enhanced. The optimized algorithm can still realize effective reconstruction within the added random error range of 0% to 7%, among which the Gaussian smoothing method presents a better optimization effect, and a reliable theoretical basis is provided for the high-precision two-dimensional reconstruction of complex combustion fields.