Title: 指紋影像辨識和影像壓縮及指紋在碎形上的分析
Identification, Compression and Fractal Analysis of Fingerprint Images
Authors: 林崇仰
Lin, Chung-Yang
吳炳飛
Wu Bing-Fei
電控工程研究所
Keywords: 指紋;離散時間小波轉換;小波尺度量化;零樹編碼;算術編碼;碎形;Fingerprint;Discrete Wavelet Transformation;Wavelet Scalar Quantization;Zerotree Coding;Arithmetic Coding;Fractal
Issue Date: 1997
Abstract: 本篇論文主要是敘述指紋在傳統辨識的方法,其中包含了取像單元、 特 徵抽取單元、資料庫單元和比對單元等四個單元。並以此辨識系統為 分 析工具,比較兩種影像壓縮方式,一種為目前美國聯邦調查局所接受 的 小波尺度量化(Wavelet Scalar Ouantization; WSQ)的壓縮方式,另 一 種為零樹加算術編碼(Zerotree and Arithmetic Coding; ZTR+AC )的 壓縮方式,個別在壓縮效果及辨識率上的表現,實驗結果顯示後者比前者 好。 此外,利用在進行WSQ或ZTR+AC的影像壓縮方法時所作的離散小波 轉換處理, 取出經離散小波轉換切割一層後低頻部分的圖形,用此和原 來指紋相似但 大小只有一半的灰階圖形,來作指紋辨識率的測試,並依 據辨識率和處理 時間的不同將辨識系統分為三級。最後,應用碎形理論 於指紋上,找出和 碎形特性相關的特徵,並利用此特徵將指紋進行粗分 類和作為辨識上的一 個參考特徵。 In this paper, we state the traditional fingerprint identification method, which it includes the capture image unit, the feature extraction unit, the database unit and the comparison unit. Utilizes this fingerprint identification system to compare the performance between two image compression methods, one is “Wavelet Scalar Ouantization” accepted by FBI and the other one is “Zerotree and Arithmetic Coding”. Experimental results are given to show the second one is better. Besides, it’s needed to take the discrete wavelet transform (DWT) in both coding algorithms. We extract the image of the low frequent coefficients from one-scale pyramidal wavelet decomposition for identification. Since the image size is half of the original image size, we can save much time in processing fingerprint images. Finally, uses the fractal theory to analysis fingerprints and finds the relative features for identification or classification.
URI: http://140.113.39.130/cdrfb3/record/nctu/#NT860591045
http://hdl.handle.net/11536/63225
Appears in Collections:Thesis