Because of the forensic, civilian and commercial needs of individual identification, fingerprints are popularly used for individual identification in a biometric system widespread. In this thesis, we reviewd the process of a minutiae-based automatic fingerprint identification system. We put emphasis on how to store the enrolled fingerprint images to make the following matching process efficiently. Our system is tested on fingerprint images of Rindex28 database to describe the information about feature points extracted from an input fingerprint image in a computer system of Pentium 4 with 3.07GHz CPU and 1.5GB RAM running Windows XP. We report a detailed statistics of a fingerprint image, and the summary statistics for the image database Rindex28 including 112 images.