Minutiae pattern remains a widely used representation of a fingerprint. The research on minutiae matching never stops due to its complexity and intractability. In this paper, an efficient fingerprint minutiae matching algorithm is proposed. To obtain reliable reference minutiae pairs, the bank of coordinate systems is introduced. The coordinate systems bank is derived from the original minutiae features and applied to get more useful information about the minutiae. To improve the accuracy of minutiae matching, a global optimum alignment approach is developed, which is targeted on the alignment of the set of reference minutiae pairs. Experimental results show that this algorithm achieves excellent performance with high matching speed and high matching reliability.