用於非正向虹膜取像之可操控傾斜式影像感測器

Steerable Skew Image Sensor for Non-orthogonal View Iris Imaging

陳侃如(Kan-Ru Chen)

指導教授 ： 石勝文

國立暨南國際大學/科技學院/資訊工程學系/碩士(2008年)

https://doi.org/10.6837/NCNU.2008.00311

摘要

生物辨識技術已經被廣泛的研究，其中虹膜辨識技術更是極受歡迎的研究主題，許多
先前的研究奠定了虹膜辨識的基礎，研究各種虹膜定位與特徵抽取的方法，無非是要提昇虹
膜辨識的準確度。而針對靜態影像資料的分析，雖然可能已經有極高的辨識率，但是拿到人
機互動環境下卻不一定能表現的一樣好，因為實際應用的互動環境下總是帶來新考驗。本
論文即是探討要如何不需讓使用者配合仍能取得高品質的虹膜影像。使用者不需要是有經
驗的使用者甚至不需移動即可自動拍到清晰的虹膜影像供虹膜辨識使用。在這個研究中我
們設計了一個可操控傾斜式影像感測器，裝置在三個自由度的機構上面，可以自由操控前
後位置以及旋轉角度。利用自動對焦演算法分析影像後移動影像感測器到最佳位置，取得
清晰影像。最後的實驗分析了各種不同對焦演算法的特性，並且展現了可自由旋轉的影像
感測器可以有效的改善影像清晰度。

關鍵字

虹膜辨識 ； 自動對焦 ； 影像感測器 ； 非正向虹膜取像

並列摘要

Biometric te?{niques have been studied extensively and iris recognition has become one
of the most popular resear?{ topics. Prior-resear?{es have focused on the iris localization
and feature extraction methods to increase the recognition rate whi?{ have become the base
of current iris recognition systems. Although the methods developed based on a static iris
database usually can a?{ieve very high recognition rate, the may not perform as well in a
dynamic environment as the static environment. In a real application, the dynamic input video
always bring up new ?{allenges. ?勇s resear?{ is focused on how to acquire high quality
images for iris recognition with less user cooperation. ?前refore, users neither need to be
experienced in using this system nor have to adjust their head pose to input their iris image.
In this study we developed a steerable skew image sensor a?|a?{ed to a three degrees of freedom
me?{anism that controls the position and rotation of the focus plane. By using auto-focusing
te?{niques, we can move the image sensor to the best focused position to acquire a high quality
iris image. In the experiments, we analyzed the propriety of various auto-focusing algorithms
and showed that the steerable skew image sensor can effectively enhance the image quality.

並列關鍵字

iris recognition ； auto-focusing ； image sensor ； non-orthogonal view iris imaging

參考文獻

[1] J. Daugman, “Uncertainty relation for resolution in space, spatial frequency, and orientation

optimized by two-dimensional visual cortical filters,” Journal of the Optical Society

of America, vol. 2, no. 7, pp. 1160–1169, 1985.

[2] J. Daugman, “Complete discrete 2-d gabor transforms by neural networks for image analysis

and compression,” IEEE Trans. on Signal Processing, vol. 36, no. 7, pp. 1169–1179,