Abstract: Star cluster features are discussed. The algorithm of Mean Shift is analyzed. The idea that distributes stars in the location field of view to clusters is proposed by iterative operation using the characteristic of star pair angular separations according to the star positions in the inertial coordinate system. The optimal selection algorithm of guide stars based on star clustering is designed. Its detailed implement procedures are introduced completely. The guide star optimal selection experiment in visible band by using SAO star catalog as the original star data is fulfilled. It proves that the proposed algorithm has the virtue of simple calculation and easy realization. The obtained guide star distribution is more uniform than the MIPS algorithm and the grid algorithm.