Please use this identifier to cite or link to this item:
標題: 以形狀和紋理相似性比對之影像擷取系統
Image Retrieval By Shape and Texture Similarity
作者: 戴紹國
Dai, Shau-Kuo
關鍵字: CBIR
image database
spatial relation
出版社: 應用數學系
摘要: 在這篇論文中, 我們提出新的方法來設計以形狀和紋理之特徵做為比對標準之影像擷取資訊系統。 在形狀相似的影像擷取中, 物件的形狀是以CPA-strings來作為索引。在第一階段,我們提出group bounding的新概念與機制來縮小資料庫的搜尋範圍, 因而加快整個查詢速度。在第二階段, 我們採用Weighted Levensthein Distance來度量CPA-strings間的距離, 若物件的距離小於我們所設定的臨界值即作為擷取的輸出結果。 在紋理相似的影像擷取方面, 我們提出了新的紋理特徵表示法, 稱為Composite Sub-band Gradient vector(CSG-vector)。實驗證明CSG-vector具有很強的紋理區分能力。為了加速影像擷取, 我們亦提出一種新的signature, 稱為EDP-string, 以及其相對應的快速配對方法, 使得資料庫中紋理差異很大的影像能夠在第一階段即被快速的篩選掉, 以提高影像的整體擷取效率。 最後, 我們針對由多個不同紋理區塊所構成的影像提出一個結合紋理特徵和區塊間空間關係的影像擷取方法, 讓系統根據使用者所選擇的紋理區塊, 抽取其CSG-vector和其空間關係來擷取最相似的影像做為輸出結果, 實驗證實這種方法不但有效率而且和人類的視覺感官一致。
In this dissertation, We propose new methods for the design of a contentbased image retrieval system based on shape, texture and compound features. In shape similarity retrieval, the shapes are indexed by CPA-strings. When a query shape is submitted to the system, it is converted to a CPAstring from which both the lower and upper bounds of the locations of the potentially matched shapes are computed. This will restrict the search space to a reasonable small proportion of the whole database. At the second stage, an angle sequence matching algorithm is invoked to compute the Weighted Levensthein Distance between the query CPA-string and the selected database CPA-strings. The shapes that have distances less than a given threshold are ‾nally retrieved. In texture similarity retrieval, the Composite Sub-band Gradient Vector (CSG-Vector) and the Energy Distribution Pattern String (EDB-String) are generated from the sub-images of a wavelet decomposition of the original image. A fuzzy matching mechanism based on EDB-Strings serves as a ‾lter to quickly remove undesired images in the database from further consideration. The images passing the ‾lter will be compared with the query image based on CSG-Vectors which are extremely powerful for discriminating detailed textures. Finally, we present a method of image retrieval based on spatial relationships between regions as well as the texture property within each region. Experimental results show that this approach is very e®ective in terms of having consistency with human visual system.
Appears in Collections:應用數學系所



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.