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

Generalized discriminant analysis in...

Yu, Jie.

FindBook

Google Book

Amazon

博客來

Generalized discriminant analysis in content-based image retrieval.

紀錄類型:	書目-語言資料,印刷品 : Monograph/item
正題名/作者:	Generalized discriminant analysis in content-based image retrieval./
作者:	Yu, Jie.
面頁冊數:	120 p.
附註:	Adviser: Qi Tian.
Contained By:	Dissertation Abstracts International68-07B.
標題:	Computer Science. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3273890
ISBN:	9780549141266

Generalized discriminant analysis in content-based image retrieval.
Yu, Jie.

Generalized discriminant analysis in content-based image retrieval. - 120 p.

Adviser: Qi Tian.

Thesis (Ph.D.)--The University of Texas at San Antonio, 2007.

Content-based Image Retrieval (CBIR) is an important computer vision application that automatically retrieves images of user interest from large database. It faces several challenges such as semantic gap, small sample size and high dimensionality. In this dissertation, novel discriminant analysis methods are generalized from the state-of-the-art techniques, such as Linear Discriminant Analysis (LDA), Principal Component Analysis (PCA) and Biased Discriminant Analysis (BDA), to develop robust and efficient CBIR systems. These methods alleviate the problems from the following aspects: Semantic Subspace Projection (SSP) models the data as lying on subspaces instead of from global Gaussian distributions as in traditional techniques such as LDA and PCA. By using semantic dissimilarity information, it bridges the gap between high-level semantic concepts and low-level features. Hybrid Discriminant Analysis (HDA) unifies LDA and PCA in a parameterized framework so that it can capture the both descriptive and discriminant information and is robust to the small sample set problem. Adaptive Discriminant Projection (ADP) combines the strength of LDA and BDA by controlling the scatterness and clusterness among samples from different classes, which allows it to handle imbalanced data sets adaptively. Interactive boosting (i.Boost) is further explored to enhance ADP by incorporating user feedback in the learning process. Novel distance measures are proposed to select the best feature elements and their corresponding distance functions automatically, which leads to more accurate similarity estimation in CBIR applications. Extensive experiments have been conducted to evaluate the performance of the proposed methods. The results have shown the superior performance of these methods over the state-of-the-art techniques.

ISBN: 9780549141266Subjects--Topical Terms:

626642
Computer Science.

Generalized discriminant analysis in content-based image retrieval.
LDR:02681nam 2200265 a 45 001 940667
005 20110518
008 110518s2007 ||||||||||||||||| ||eng d
020 $a 9780549141266
035 $a (UMI)AAI3273890
035 $a AAI3273890
040 $a UMI $c UMI
100 1 $a Yu, Jie. $3 1263053
245 1 0 $a Generalized discriminant analysis in content-based image retrieval.
300 $a 120 p.
500 $a Adviser: Qi Tian.
500 $a Source: Dissertation Abstracts International, Volume: 68-07, Section: B, page: 4615.
502 $a Thesis (Ph.D.)--The University of Texas at San Antonio, 2007.
520 $a Content-based Image Retrieval (CBIR) is an important computer vision application that automatically retrieves images of user interest from large database. It faces several challenges such as semantic gap, small sample size and high dimensionality. In this dissertation, novel discriminant analysis methods are generalized from the state-of-the-art techniques, such as Linear Discriminant Analysis (LDA), Principal Component Analysis (PCA) and Biased Discriminant Analysis (BDA), to develop robust and efficient CBIR systems. These methods alleviate the problems from the following aspects: Semantic Subspace Projection (SSP) models the data as lying on subspaces instead of from global Gaussian distributions as in traditional techniques such as LDA and PCA. By using semantic dissimilarity information, it bridges the gap between high-level semantic concepts and low-level features. Hybrid Discriminant Analysis (HDA) unifies LDA and PCA in a parameterized framework so that it can capture the both descriptive and discriminant information and is robust to the small sample set problem. Adaptive Discriminant Projection (ADP) combines the strength of LDA and BDA by controlling the scatterness and clusterness among samples from different classes, which allows it to handle imbalanced data sets adaptively. Interactive boosting (i.Boost) is further explored to enhance ADP by incorporating user feedback in the learning process. Novel distance measures are proposed to select the best feature elements and their corresponding distance functions automatically, which leads to more accurate similarity estimation in CBIR applications. Extensive experiments have been conducted to evaluate the performance of the proposed methods. The results have shown the superior performance of these methods over the state-of-the-art techniques.
590 $a School code: 1283.
650 4 $a Computer Science. $3 626642
690 $a 0984
710 2 $a The University of Texas at San Antonio. $3 1252033
773 0 $t Dissertation Abstracts International $g 68-07B.
790 $a 1283
790 1 0 $a Tian, Qi, $e advisor
791 $a Ph.D.
792 $a 2007
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3273890