The method includes: obtaining features of L key points of a first fingerprint image; determining G pairs of key points based on the L key points of the first fingerprint image and L key points of a first fingerprint template; sorting the G pairs of key points based on the first distance parameter corresponding to each pair of key points, and determining N pairs of key points from the G pairs of key points that are sorted; determining a first rotation matrix using the N pairs of key points, and rotating the first fingerprint image based on the first rotation matrix to obtain a second fingerprint image, where the second fingerprint image is in a same direction as the first fingerprint template; and performing similarity matching using the second fingerprint image and the first fingerprint template.