基於色彩資訊特徵之虹膜辨識

Iris Recognition based on Color Information Features

鄭仁和(Ren-He Jeng)

指導教授 ： 陳文雄

國立暨南國際大學/科技學院/電機工程學系/碩士(2009年)

https://doi.org/10.6837/NCNU.2009.00043

摘要

生物辨識技術中人眼虹膜辨識技術，大多著重於虹膜結構即為虹膜紋理影像研究和討論，鮮少人眼虹膜辨識研究採用虹膜的色彩資訊。因此，本論文將討論人眼虹膜色彩資訊特徵是否具備身份辨識能力。
本論文人眼虹膜資料庫採用葡萄牙內貝拉大學計算機科學演算法暨影像分析學群（Soft Computing and Image Analysis Group, Department of Computer Science, University of Beira Interior）所提供 UBIRIS 虹膜資料庫，使用第一大類241個類別1205張人眼擷取影像。生物辨識流程可分為三個步驟，「感興趣人眼虹膜區域擷取」、「特徵萃取」和「相似度比較」。特徵萃取步驟進行三種成份實驗分別為「結構」、「空間」和「成份比例」；結構成份採用二維伴隨矩陣、三維伴隨矩陣統計相素對關係。空間成份採用K-means與中心分裂演算法找出空間質心位置。比例和成份中採用圖案伴隨矩陣、二維線性識別分析、二維主成份分析、K-means直方圖統計、色彩矩、三維色彩伴隨矩陣和區域色彩成份比例關係方式。實驗色彩資訊包括Texture、Gray、RGB、YIQ、YCbCr、CIELab和以K-means演算法進行色彩量化影像進行特徵萃取。於相似度比較分別使用歐基里德距離、漢明距離、餘弦相似度和豪斯多夫距離。實驗結果由本論文所提區域色彩成份比例關係方式具有最佳辨識結果，最佳實驗結果等錯誤率為1.1%，驗證人眼虹膜色彩特徵資訊是獨特且有用的。

關鍵字

虹膜 ； 辨識 ； 色彩 ； 結構 ； 空間 ； 比例和成份

並列摘要

In biometric of iris recognition technology, the spatial patterns have been studied and recognized electively for several years. The iris recognition researches most attention on gray-level image. This paper delves into the color distribution of iris recognition has rich information enough to authenticate personal identity. This paper delved into the problem of whether iris color has enough information to verify personal identity or not.
This paper proposed an iris recognition system that tested on the UBIRIS database which includes 1205 images from 241 classes. The iris recognition system consists of iris localization, feature extraction, and pattern matching. In this paper, we investigate some color features information described by the following three forms: structure, space and proportion. The iris recognition experiments utilize different image color descriptors, such as texture, gray, RGB, YIQ, YCbCr, CIELab, and other quantization of color space. In order to observe the structure features, we adopt 2D co-occurrence matrix (2D-CM) and 3D co-occurrence matrix (3D-CM). The analysis of space features uses K-means algorithm and centriod splitting algorithm to find cluster centers. The analysis of proportion features, we carry out the experiments extensively with a number of feature extraction methods including color moment, color probabilities distributions module (CPDM), two-dimensional principal components analysis (2D-PCA), two-dimensional linear discriminate analysis (2D-LDA), K-means Histogram, three-dimensional color level co-occurrence matrix (3D-CLCM), motif co-occurrence matrix (MCM), and we proposed of local color proportion descriptor (LCPD) that encode combined statistic measure with Peano scanning. According to the analysis and comparison, a significant summary is given. The feature of color information is useful and uniquely.

並列關鍵字

Iris ； Recognition ； Color ； Structure ； Space ； Proportion

參考文獻

[1] J. G. Daugman, “How iris recognition works,” IEEE Trans. on Circuits and Systems for Video Technology, vol. 14, no.1, pp. 21-30, Jan. 2004.

[2] J. G. Daugman, “Biometric personal identification system based on iris analysis,” United States Patent, no. 5,291,560, Mar. 1994.

[3] J. G. Daugman, “High confidence visual recognition of persons by a test of statistical independence,” IEEE Trans. on Pattern Analysis and Machine Intelligence, vol. 15, no. 11, pp. 1148-1161, Nov. 1993.

[4] R. P. Wildes, “Iris recognition: an emerging biometric technology,” Proc. of the IEEE, vol. 85, no. 9, pp. 1348-1363, Sep. 1997.

[5] R. P. Wildes, “Automated, non-invasive iris recognition system and method,” United States Patent, no: 5572596, David Sarnoff Research Center Inc., Nov. 1996.