Abstract: Algorithm of projection reconstruction from fewer views is a puzzle in the field of image reconstruction, genetic algorithm is one of these algorithms applied in this field, but there are some disadvantages in it. In order to overcome these disadvantages existing in optimization process, such as premature convergence, simulated annealing was introduced into genetic algorithm and a new algorithm was created. By selecting test function, an optimized model for incomplete projection image reconstruction was built based on this algorithm. The computer simulation results indicate that this new algorithm could achieve accurate image reconstruction with data obtained from evenly distributed 12 projection angles of 15°.