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

Performance prediction for biometric...

Wang, Rong.

FindBook

Google Book

Amazon

博客來

Performance prediction for biometrics recognition systems.

紀錄類型:	書目-語言資料,印刷品 : Monograph/item
正題名/作者:	Performance prediction for biometrics recognition systems./
作者:	Wang, Rong.
面頁冊數:	141 p.
附註:	Adviser: Bir Bhanu.
Contained By:	Dissertation Abstracts International68-09B.
標題:	Engineering, Electronics and Electrical. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3281698
ISBN:	9780549226956

Performance prediction for biometrics recognition systems.
Wang, Rong.

Performance prediction for biometrics recognition systems. - 141 p.

Adviser: Bir Bhanu.

Thesis (Ph.D.)--University of California, Riverside, 2007.

Performance prediction is a fundamental problem for a biometrics recognition system. In this thesis, we address the biometrics performance prediction for large populations and the performance prediction for multisensor fusion systems.

ISBN: 9780549226956Subjects--Topical Terms:

626636
Engineering, Electronics and Electrical.

Performance prediction for biometrics recognition systems.
LDR:03474nam 2200289 a 45 001 958205
005 20110704
008 110704s2007 ||||||||||||||||| ||eng d
020 $a 9780549226956
035 $a (UMI)AAI3281698
035 $a AAI3281698
040 $a UMI $c UMI
100 1 $a Wang, Rong. $3 1273485
245 1 0 $a Performance prediction for biometrics recognition systems.
300 $a 141 p.
500 $a Adviser: Bir Bhanu.
500 $a Source: Dissertation Abstracts International, Volume: 68-09, Section: B, page: 6203.
502 $a Thesis (Ph.D.)--University of California, Riverside, 2007.
520 $a Performance prediction is a fundamental problem for a biometrics recognition system. In this thesis, we address the biometrics performance prediction for large populations and the performance prediction for multisensor fusion systems.
520 $a We present a binomial model to predict a large population performance based on a small gallery. Using the match score and non-match score obtained from the small gallery, we estimate distributions of the match score and the non-match score for a large population. We use these distributions in the prediction model that follows the binomial distribution to predict the performance on a large population. In order to model the distortion happened in large populations, we propose a two-dimensional model that combines a hypergeometric probability distribution model with a binomial model to predict a large population performance from a small gallery. Our distortion model includes feature uncertainty, feature occlusion, and feature clutter. By an iterative learning process, we find the optimal size of a small gallery. We use the Chernoff inequality and the Chebychev inequality to determine the small gallery size in theory which is related to the margin of error and the confidence interval. We find the upper bound and a good lower bound for predicting recognition performance on a large population.
520 $a We present two theoretical approaches to predict the sensor fusion performance that allow us to select the optimal sensor combination. We assume that the match score and the non-match score distributions are mixture of Gaussians. We novelly apply the Neyman-Pearson theory to predict the sensor fusion performance. We develop a measurement which considers not only the mean and variance but also the skewness of the log-likelihood ratio of the fusion system similarity scores as our discriminability measurement to find the optimal fusion combination instead of doing the Brute-Force experiments. Based on the theory that the area under the ROC curve (AU ROC) can be used to evaluate the recognition system performance, we present a prediction model which decomposes the AU ROC of the fusion system to a set of AU ROCs which are obtained from the combinations of the components from the match score distribution and the non-match score distribution. Then, we use an explicit phi transformation that maps a ROC curve to a straight line in 2-D space whose axes are related to the false alarm rate (FAR) and the Hit rate. In steading of computing the AU ROC, we derive a metric to evaluate the sensor fusion performance and find the optimal sensor combination.
590 $a School code: 0032.
650 4 $a Engineering, Electronics and Electrical. $3 626636
690 $a 0544
710 2 $a University of California, Riverside. $3 791122
773 0 $t Dissertation Abstracts International $g 68-09B.
790 $a 0032
790 1 0 $a Bhanu, Bir, $e advisor
791 $a Ph.D.
792 $a 2007
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3281698