Projection decomposition of dual-energy computed tomography (CT) is a key step for dual-energy CT pre-reconstruction algorithm. The selection of iterative initial values has great influence on the convergence of the iteration when an iterative algorithm is use to solve the dual-energy projection integral equations. To solve the problem of selecting initial values in solving dual-energy projection integral equations, an optimized iterative algorithm is proposed. For calculating projection lookup tables of dual-energy basis materials and calibrating the system, a certain step length is set based on numerical ranges of high-energy and low-energy projection data. Thus the corresponding basis materials projection can be solved according to the order of the projection values. The convergent result of last time is taken as the initial value of next iteration. The results show that the proposed algorithm can effectively improve the convergence of the iteration and the efficiency of calibration of the duel-energy CT system, which proves the validity of the algorithm.