基於紋理分析法－應用於虹膜識別之研究

• Main Authors: Huang Min-Yu, 黃敏昱
• Other Authors: 張劍平
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/97853459416306361864

博士 === 國防大學理工學院 === 國防科學研究所 === 98 === Biometric is a science of research for automatic recognition of individuals based on physiological (face, fingerprint, iris, palm-prints, DNA) and behavioral (gait, graphology, voiceprint) characteristics. Iris recognition has many advantages, including that it is stable, non-intrusive and difficult to be changed. Iris recognition has been a research hot-spot in personal identification field of biometrics. We present a whole iris recognition system, but particularly focus on the image quality assessment and propose Local Binary Pattern (LBP), Modified Empirical Mode Decomposition (MEMD) and Improved Empirical Mode Decomposition (IEMD) to extract features for iris recognition. Experiments are conducted on the public and freely available iris images from the CASIA and UBIRIS databases. To evaluate the outcomes, three different similarity measures are used in the experiment. The experimental results show that the presented schemes achieve promising performance by those three measures. Therefore three proposed methods are feasible for iris recognition and LBP, MEMD, and IEMD are suitable for iris feature extraction.