Based on minutiae points algorithm, this paper proposes a fingerprint recognition technique to match one fingerprint among various fingerprint images accurately. First, in order to raise quality of blurred and broken images, we implement the preprocess using enhancement and Gabor filter. Then two kinds of minutiae points are extracted from the print and validated. Finally, during the matching step, points are matched using Ransac under an affine transformation model. After testing upon database FVC2002 and FVC2004, we prove that our method is robust and accurate with 85% accuracy.