Abstract: The temperature distribution of space objects can reflect in-orbit state. Ground-based and long-focal length imaging of space objects in infrared band provides information from which temperature distribution maps can be determined. The methods of the traditional temperature estimation are often single-band radiance and dual-band color thermometry. Uncertainties in the imaging process will reduce the traditional single-band radiance and dual-band color thermometry forward calculation accuracy. Such uncertainties include target emissivity, atmospheric transmittance, the earth's heat radiation, etc. In this paper, multi-band Bayesian estimation function model relating the measured number of electrons was established, the true temperature of space target could be more accurately solved. The model provided an estimation theoretic framework for developing optimal estimators and calculating Cramer-Rao lower bounds. The Cramer-Rao bounds of target parameter estimation function was derived, which could predict the accuracy of solving of the temperature and emissivity. Finally, the simulation and analysis of the algorithm was performed.