東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

Efficient biometric authentication b...

Mitra, Sinjini.

FindBook

Google Book

Amazon

博客來

Efficient biometric authentication based on statistical models.

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Efficient biometric authentication based on statistical models./
作者:	Mitra, Sinjini.
面頁冊數:	227 p.
附註:	Source: Dissertation Abstracts International, Volume: 66-10, Section: B, page: 5480.
Contained By:	Dissertation Abstracts International66-10B.
標題:	Statistics. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3190068
ISBN:	9780542349683

Efficient biometric authentication based on statistical models.
Mitra, Sinjini.

Efficient biometric authentication based on statistical models. - 227 p.

Source: Dissertation Abstracts International, Volume: 66-10, Section: B, page: 5480.

Thesis (Ph.D.)--Carnegie Mellon University, 2005.

In the modern electronic information age, there is an ever-growing need to authenticate and identify individuals for ensuring the security of systems. Based on a person's unique physiological traits (which is distinct from the more traditional statistical field of biometry), biometric authentication is more reliable than traditional PINs and ID cards. The field of biometrics has grown exponentially in recent years (especially after September 11), and the recent practice of recording face and fingerprint of foreign passengers at all U.S. shows the growing importance of biometrics in U.S. homeland security.

ISBN: 9780542349683Subjects--Topical Terms:

517247
Statistics.

Efficient biometric authentication based on statistical models.
LDR:03431nmm 2200337 4500 001 1825997
005 20061206120642.5
008 130610s2005 eng d
020 $a 9780542349683
035 $a (UnM)AAI3190068
035 $a AAI3190068
040 $a UnM $c UnM
100 1 $a Mitra, Sinjini. $3 1914970
245 1 0 $a Efficient biometric authentication based on statistical models.
300 $a 227 p.
500 $a Source: Dissertation Abstracts International, Volume: 66-10, Section: B, page: 5480.
500 $a Advisers: Stephen E. Fienberg; Anthony Brockwell.
502 $a Thesis (Ph.D.)--Carnegie Mellon University, 2005.
520 $a In the modern electronic information age, there is an ever-growing need to authenticate and identify individuals for ensuring the security of systems. Based on a person's unique physiological traits (which is distinct from the more traditional statistical field of biometry), biometric authentication is more reliable than traditional PINs and ID cards. The field of biometrics has grown exponentially in recent years (especially after September 11), and the recent practice of recording face and fingerprint of foreign passengers at all U.S. shows the growing importance of biometrics in U.S. homeland security.
520 $a This thesis focuses on establishing the role of statistical models in developing efficient face authentication systems and also in evaluating the accuracy of proposed methods. We propose a model-based system in the frequency domain by exploiting the well-known significance of the phase component in face identification, after observing that spatial models are inadequate for the purpose. We compare the performance of our system with an existing technique based on a linear filter called the Minimum Average Correlation Energy (MACE) filter, which despite its success suffers from a number of drawbacks owing to its non-model-based framework.
520 $a We next explore the role of facial asymmetry in devising efficient authentication tools using both spatial and frequency domain representation. Furthermore, a comparison of the two representations establishes that the frequency domain measures are more robust to intra-personal distortions. We explore a potential connection with phase and perform an extensive feature analysis performed to determine the specific facial regions that contribute to the recognition tasks. We also explore application to video data and modeling options.
520 $a The thesis also introduces a general statistical framework based on random effects models for evaluating the performance of any biometric system on large-scale real-world databases. This framework helps in predicting performance of a system on unknown databases based on test-bed data. When we apply this methodology to the MACE and the mixture model-based systems, the latter outperforms the former in terms of predictive performance. The method is also applied to the asymmetry-based system.
520 $a Finally, we investigate generalizing the scope of these methodologies to other biometrics with applications to fingerprint data.
590 $a School code: 0041.
650 4 $a Statistics. $3 517247
650 4 $a Biology, Biostatistics. $3 1018416
690 $a 0463
690 $a 0308
710 2 0 $a Carnegie Mellon University. $3 1018096
773 0 $t Dissertation Abstracts International $g 66-10B.
790 1 0 $a Fienberg, Stephen E., $e advisor
790 1 0 $a Brockwell, Anthony, $e advisor
790 $a 0041
791 $a Ph.D.
792 $a 2005
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3190068