DC 欄位 |
值 |
語言 |
DC.contributor | 資訊管理學系 | zh_TW |
DC.creator | 廖雅蓮 | zh_TW |
DC.creator | Ya-Lien Liao | en_US |
dc.date.accessioned | 2010-7-16T07:39:07Z | |
dc.date.available | 2010-7-16T07:39:07Z | |
dc.date.issued | 2010 | |
dc.identifier.uri | http://ir.lib.ncu.edu.tw:88/thesis/view_etd.asp?URN=974203011 | |
dc.contributor.department | 資訊管理學系 | zh_TW |
DC.description | 國立中央大學 | zh_TW |
DC.description | National Central University | en_US |
dc.description.abstract | 隨著電腦網路與資訊科技蓬勃發展,影像自動命名與檢索的議題應運而生,如何提供使用者準確及有效率的影像檢索成為探索的目標。近年來以影像內容特徵為基礎的影像檢索方法蓬勃發展。然而,直接使用影像原始低階特徵的分類準確率偏低,造成影像經常被命名了不恰當的關鍵字。本論文提出一種新的特徵表示法達成優於使用原始低階特徵的分類準確率,稱為Meta-Features 表示法。本論文使用Corel 資料集來驗證Meta-features 能夠達成較好的分類成效。為了充分驗證,實驗採用三種知名的分類器來進行比較,分別是Support Vector Machines (SVM)、k-Nearest Neighbor (k-NN)與naïve Bayes 分類器。
| zh_TW |
dc.description.abstract | This thesis proposes a feature representation approach namely meta-feature representation for automatic image annotation. Automatic image annotation technology aims to provide an efficient and effective searching environment for users to query images by keywords, which solves the limitation of Content-Based Image Retrieval (CBIR) that low-level image features are only extracted and used for similarity search. It is the fact that low-level image features do not directly correspond to high-level concepts of users. This causes many incorrect keyword assignments to images since a certain number of semantically similarn images have dissimilar low-level features and semantically dissimilar images have similar low-level features. The
meta-features proposed in this thesis are extracted from the original low-level image features by nine transformation formulas to improve annotation accuracy. We use the Corel dataset for the experiments to show the performance improvement of image annotation based on the meta-features. In particular, for classifier design, this thesis considers three well-known and popular classifiers for image annotation. They are the k-Nearest Neighbor (k-NN) classifier, Support Vector Machines (SVM), and the naïve Bayes classifier.
| en_US |
DC.subject | 圖片特徵表示法 | zh_TW |
DC.subject | 影像檢索 | zh_TW |
DC.subject | 圖片命名 | zh_TW |
DC.subject | 圖片分類 | zh_TW |
DC.subject | feature representation | en_US |
DC.subject | image retrieval | en_US |
DC.subject | image annotation | en_US |
DC.subject | image classification | en_US |
DC.title | A Meta-Feature Representation Approach toImage Annotation | en_US |
dc.language.iso | en_US | en_US |
DC.type | 博碩士論文 | zh_TW |
DC.type | thesis | en_US |
DC.publisher | National Central University | en_US |