語系:
繁體中文
English
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
Contributions to content-based image...
~
Xie, Hua.
FindBook
Google Book
Amazon
博客來
Contributions to content-based image retrieval.
紀錄類型:
書目-電子資源 : Monograph/item
正題名/作者:
Contributions to content-based image retrieval./
作者:
Xie, Hua.
面頁冊數:
162 p.
附註:
Source: Dissertation Abstracts International, Volume: 66-11, Section: B, page: 6185.
Contained By:
Dissertation Abstracts International66-11B.
標題:
Engineering, Electronics and Electrical. -
電子資源:
http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3196914
ISBN:
9780542427862
Contributions to content-based image retrieval.
Xie, Hua.
Contributions to content-based image retrieval.
- 162 p.
Source: Dissertation Abstracts International, Volume: 66-11, Section: B, page: 6185.
Thesis (Ph.D.)--University of Southern California, 2005.
Due to the proliferation of multimedia information over the internet, users are confronted with large amounts of content from many sources around the world. Content-based retrieval system have been proposed to automatically annotate and index multimedia information based on their audio/visual contents instead of manually-entered text keywords. In this thesis, we investigate two major topics related to content-based retrieval.
ISBN: 9780542427862Subjects--Topical Terms:
626636
Engineering, Electronics and Electrical.
Contributions to content-based image retrieval.
LDR
:03287nmm 2200301 4500
001
1822169
005
20061127121649.5
008
130610s2005 eng d
020
$a
9780542427862
035
$a
(UnM)AAI3196914
035
$a
AAI3196914
040
$a
UnM
$c
UnM
100
1
$a
Xie, Hua.
$3
1911321
245
1 0
$a
Contributions to content-based image retrieval.
300
$a
162 p.
500
$a
Source: Dissertation Abstracts International, Volume: 66-11, Section: B, page: 6185.
500
$a
Adviser: Antonio Ortega.
502
$a
Thesis (Ph.D.)--University of Southern California, 2005.
520
$a
Due to the proliferation of multimedia information over the internet, users are confronted with large amounts of content from many sources around the world. Content-based retrieval system have been proposed to automatically annotate and index multimedia information based on their audio/visual contents instead of manually-entered text keywords. In this thesis, we investigate two major topics related to content-based retrieval.
520
$a
First, we propsose and analyze effecient compression techniques for distributed image retrieval systems. The first technique we design is a classified quantization system. A partial classification is first performed before compressing the data so that we are able to capture the special characteristics of the classes that are relevant to content-based retrieval. The pre-classifier and the quantization parameters for each class are jointly searched based on a rate-distortion-complexity optimization framework. Substantial improvement in terms of retrieval performance vs. bit rate, is achieved using the proposed compression scheme as compared to standard encoding.
520
$a
The second technique we consider is to use linear discriminant analysis for transform coding in distributed image classification/retrieval systems. We examine the optimal transform which compacts the class discrimination information into the lowest dimensional space, and propose a greedy bit allocation algorithm to minimize the loss in class separability due to quantization. We analyze the relations between proposed transform coding and Likelihood Ratio Quantization, and develope high rate analysis for certain classes of Gaussian distributions.
520
$a
The second topic addresses relevance feedback, a critical component for content-based retrieval systems. It has been shown that support vector machines (SVMs) can be used to conduct effective relevance feedback. In this work, we propose an approach to derive a novel information divergence based kernel given the user's preference. Our proposed kernel function naturally takes into account the statistics of the data that is available during relevance feedback. Experiments show that the new kernel achieves significantly higher (about 17%) retrieval accuracy than the standard radial basis function (RBF) kernel, and can thus become a valid alternative to traditional kernels for SVM-based active learning in relevance feedback applications.
590
$a
School code: 0208.
650
4
$a
Engineering, Electronics and Electrical.
$3
626636
690
$a
0544
710
2 0
$a
University of Southern California.
$3
700129
773
0
$t
Dissertation Abstracts International
$g
66-11B.
790
1 0
$a
Ortega, Antonio,
$e
advisor
790
$a
0208
791
$a
Ph.D.
792
$a
2005
856
4 0
$u
http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3196914
筆 0 讀者評論
館藏地:
全部
電子資源
出版年:
卷號:
館藏
1 筆 • 頁數 1 •
1
條碼號
典藏地名稱
館藏流通類別
資料類型
索書號
使用類型
借閱狀態
預約狀態
備註欄
附件
W9213032
電子資源
11.線上閱覽_V
電子書
EB
一般使用(Normal)
在架
0
1 筆 • 頁數 1 •
1
多媒體
評論
新增評論
分享你的心得
Export
取書館
處理中
...
變更密碼
登入