摘 要 數位化指紋比對的發展已有三十多年的歷史。傳統比對指紋的方式,主要可分為細微特徵點式比對與濾波器式比對兩大類。它們皆是萃取指紋紋線特徵為比對依據,將輸入的指紋與儲存在資料庫中的指紋特徵樣本進行比對。但被廣泛使用的細微特徵點比對方式在遭遇未知細微特徵點時,比對速度將大幅下降。而使用濾波器式的比對方式,若輸入的指紋有缺陷情形,則比對時很容易產生相當大的誤判率(FAR)。本論文提出一種相位式的比對演算法,使用相位差進行指紋比對,有別於傳統使用歐氏距離的比對方式,而將。振幅正規化為一。經過實際實驗結果顯示,使用相位式比對演算法具有較佳的強健性,並可有效加快指紋的比對速度與降低比對誤判率(FAR)。
Abstract The technique of digital fingerprint matching has been used for more than thirty years. Two main traditional approaches to do fingerprint matching are minutiae-based matching and filter-based matching. They were proposed to extract the features of ridges and valleys pattern and verify the incoming fingerprint with stored fingerprint database fast and accurately. But the widely used minutiae-based matching has difficulty in quickly matching two fingerprint images containing different number of unregistered minutiae points. And the filter-based matching suffered much in false acceptance rate when dealt with defective fingerprints. The proposed phase-based matching algorithm uses phase difference to do fingerprint verification instead of traditional Euclidean distance. The amplitude value is normalized into unitary value. This robust phase-matching algorithm speeds up the fingerprint matching and reduces false acceptance rate after several experiments.
為了持續優化網站功能與使用者體驗,本網站將Cookies分析技術用於網站營運、分析和個人化服務之目的。
若您繼續瀏覽本網站,即表示您同意本網站使用Cookies。