- 學位論文
複合式掌紋識別系統
Bi-feature Personal Identification using Hand Geometry and Palmprint Image Biometrics
摘要
摘 要 資訊科技的發展配合生產技術的進步，成本低廉的網路資源已隨手可得，人們期待科技的進步帶來生活品質的改善，但卻也可能成為非法活動的工具，有效而便利的安全識別系統一直是學術和產業界致力發展的目標，其中生物辨識的便利性便是眾所期待的系統之一。指紋、虹膜、聲紋、臉像、和手掌是目前較常用的特徵，其中手掌辨識的技術又分為紋路特徵、掌形比對、掌形量測以及靜脈顯影等等不同驗證方式。其中掌紋辨識則是本文研究的主題。 本文是探討一複合式掌紋識別系統，結合平台式掃描器與新的手掌水平定位架作為本系統的影像擷取設備，新的定位方式提供良好的使用彈性和直覺的使用性。本系統採用雙特徵生物辨識：包含了十八個手掌的幾何尺寸特徵和五個掌紋影像灰階特徵。在初期的評估階段可以得到各特徵對應的加權參數，在識別階段則採用兩階段式的差異性驗證法，將幾何與影像特徵結合應用。本系統藉由一系列的實驗，其間共87人完成擷取，並取得207張掌紋影像，實驗結果顯示全部的掌紋影像都能於約十秒鐘左右被正確識別。
並列摘要
In this thesis, a composite personal palmprint identification system is investigated. The horizontal palm positioning frame on a flat-bed scanner is proposed as well. This proposed orientation method provides both high flexibility and usability. Both palm geometry and image biometrics data were applied in this system. There are 18 palm dimensions and 5 grayscale histogram features are used with their weighting factors. Those factors were determined at the initial evaluation stage. At the identification stage, the dissimilarity verification combined two steps of biometrics, using geometric and image features. This system is tested by a series of experiments, during which totally 207 palmprint images from 87 persons are captured. The experiment results show that all of the palmprint images can be correctly distinguished within around 10 seconds.