基於KD-TREE的高資料量最近鄰居搜尋法

機構典藏 > College of Engineering > Graduate Institute & Department of Computer Science and Information Engineering > Thesis > Item 987654321/77462

Please use this identifier to cite or link to this item: https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/77462

Title:	基於KD-TREE的高資料量最近鄰居搜尋法
Other Titles:	KD-TREE based nearest neighbor search for large quantity data
Authors:	謝雅如;Hsieh, Ya-Ju
Contributors:	淡江大學資訊工程學系碩士班顏淑惠;Yen, Shwu-Huey
Keywords:	特徵點;KD-樹;KDA;最近鄰居搜尋;影像拼接;Feature Point;KD-Tree;Arbitrary KD-tree (KDA);Nearest Neighbor (NN);Image Stitching
Date:	2012
Issue Date:	2012-06-21 06:47:06 (UTC+8)
Abstract:	尋找事先未經訓練的最近鄰居，有許多的應用，例如: 馬賽克圖片，影像匹配，影像檢索，影像縫合。當資料量是大量的而且資料的維度高，如何有效地找到最近的鄰居就顯得非常重要。在這篇文章中，我們比較若干KD-樹的變化型以搜索最近鄰居。基於合成資料的測試，我們得出的結論是 KD樹的方法運作在有相關性資料比運作在不相連性的隨機資料好。還有，利用影像產生SIFT特徵點的測試，我們得出本文所提出的方法KDA以演算法分析其計算複雜度，比傳統的KD-Tree省下不少執行效能。最後，我們結合本文所提出的方法加以應用到影像拼接上。 Finding nearest neighbors without training in advance has many applications, such as image mosaic, image matching, image retrieval, and image stitching. When the quantity of data is huge and dimension is high, how to efficiently find the nearest neighbor (NN) becomes very important. In this article, we propose a variation of the KD-tree, Arbitrary KD-tree (KDA) which build tree without evaluate variances. Multiple KDA not only can be built efficiently it also processes an independent tree structures when data is large. Tested by extended synthetic databases and real-world SIFT data, we concluded that KDA method has advantages of satisfying accuracy performance in NN problem as well as computation efficiency.
Appears in Collections:	[Graduate Institute & Department of Computer Science and Information Engineering] Thesis

Files in This Item:

File	Size	Format
index.html	0Kb	HTML	338	View/Open

Loading...