Abstract: The laser absorption spectroscopy has a broad application prospect in the field of Aeronautics and Astronautics. In this paper, a two-dimensional gas temperature is reconstructed by the multi-wavelengths diode laser absorption spectroscopy. A measurement system for the gas temperature is designed, and the number of projections is reduced by increasing the information of the absorption spectrum. A hybrid algorithm which combines the genetic simulated annealing with the sequential quadratic programming (SQP) is used to solve the problem, which improves the efficiency of the search. A combustion model of the product including water is established, and the temperature of the combustion field is reconstructed by using 8 near-infrared absorption spectrum lines of water. The stability of the algorithm is verified by adding different proportions of random noise in the measurement information, and the results are compared with the two wavelength absorption method. The experimental results show that the reconstructions of the two methods are equivalent in the low-noise environment, but with the increase of the noise, the multi-wavelength absorption method performs more stable.