應用雲端運算於超大型影像資料庫檢索｜國立臺灣科技大學博碩士論文系統

簡易檢索 / 詳目顯示

回結果列表

研究生：	歐陽蒨 Chien Ou-Yang
論文名稱：	應用雲端運算於超大型影像資料庫檢索 A Cloud-Computing Based Very Large Scale Image Retrieval System
指導教授：	陳建中 Jiann-Jone Chen
口試委員:	吳怡樂 Yi-Leh Wu 唐政元 Cheng-Yuan Tang 何瑁鎧 Maw-Kae Hor
學位類別：	碩士 Master
系所名稱：	電資學院 - 電機工程系 Department of Electrical Engineering
論文出版年：	2014
畢業學年度：	103
語文別：	中文
論文頁數：	79
中文關鍵詞：	影像檢索、雲端運算
外文關鍵詞：	content-based image retrieval, cloud-computing, big data
相關次數：	點閱：200 下載：2
分享至:	分享至facebook 分享至twitter

查詢本校圖書館目錄查詢臺灣博碩士論文知識加值系統勘誤回報

近年來個人行動裝置普及，透過智慧型手機及平板電腦經網際網路搜尋資
料的已經成為生活的一部份，在社群網站如Facebook, Instagram, 以及Flickr 的
普及，每位使用者就如同擁有一個小型的行動影像資料庫，可以隨時取像及搜尋，
因此如何有效的從媒體大海中找到與使用者興趣相符合的影像，成為一個具潛力
的應用市場，本論文研究MPEG7 的視覺搜尋精簡描述子(Compact Descriptors for
Visual Search, CDVS)特性，運用遠端大型資料庫及運算伺服器的即時搜尋，
將檢索結果回傳至使用者行動裝置，針對這個應用我們設計遠端伺服器運用雲端
運算來達成超大型視覺資料庫搜尋的方法，以開發一個符合MPEG7 CDVS 規範
的應用系統。因應現今社會下有大量影像資料透過網路傳送及瀏覽，在大數據環
境下應該如何去處理資訊(Big Data Analysis)，本論文提出在雲端運算Hadoop 的
架構下，有效應用分散式運算來達成快速超大型影像資料庫檢索的目的，資料庫
為百萬張影像的規模，透過分散式檔案系統(Hadoop Distributed File System,
HDFS)來做資料庫的儲存及管理。在影像檢索方面，本論文採用MPEG7 所制訂
的特徵描述子中的五種描述子(色彩結構描述子、色彩可調性描述子、色彩佈局
描述子、邊緣直方圖描述子、同質紋理描述子)做為基礎，並據此提出多特徵的
正規化相關係數檢索方法，來達到良好的檢索結果。實驗結果顯示，本論文所提
出的雲端運算巨資分析架構，能有效的降低影像特徵擷取時間，所需運算時間與
未應用雲端運算的系統相較，可減少83.79%的運算時間。實驗結果亦顯示應用
雲端運算在大型影像資料庫檢索的情況下，當資料庫規模越大，所能降低的檢索
時間比例也越大，而應用MPEG7 之多種影像特徵來進行檢索的情況下，在大型
資料庫的應用架構下，所能達到的檢索率也越高。

Personal mobile devices, such as smart phone and table PC, become popular in
recent year, with the help of widespread network communication services. Social
network services, such as Facebook, Instangram, and Flicker, become our everyday
life. The personal mobile device now acts like a small image database, which means
that user can acquire images, store them and perform visual search. However, due to
limited communication and computation power, the personal mobile device has to
utilize remote powerful servers to perform big data analysis for information retrieval.
Based on MPEG7 visual descriptors, a standard of compact descriptors for visual
search (CDVS) was proposed, such that users can perform effectively perform visual
search on a very large scale image database with mobile devices, whose computing
and communication powers are limited. In response to the Big Data world, how to
effectively perform visual search become important. We proposed to utilize a cloud
computing framework, Hadoop, to perform image retrieval for large scale image
databases, and utilize Hadoop distributed file system (HDFS) to manage database and
descriptors access. For image visual descriptors, MPEG7 image descriptors, such as
color structure descriptor, color scalable descriptor, color layout descriptor, edge
histogram descriptor, and homogeneous texture descriptor are adopted to perform
image retrieval. We proposed to utilize a multi-feature normalized correlation
coefficient retrieval method to achieve high efficient image retrieval on large scale
image databases. Experiments showed that the cloud-computing based CDFS can
save feature extraction time up to 83.79%. For cloud-computing based image
retrieval, the time can be further reduced when the scale of image database became
much larger. When adopting multi-feature image retrieval, the time reduction
efficiency can also be further improved, while achieving better retrieval accuracy.

應用雲端運算於超大型影像資料庫檢索 ............................ I
摘要 .......................................................... I
ABSTRACT ..................................................... II
致謝 ........................................................ III
目錄 .......................................................... V
圖目錄 ..................................................... VIII
表目錄 ....................................................... XI
第一章 緒論 ................................................ 1
1 研究背景與動機............................................ 1
2 研究項目與方法概述........................................ 2
3 本論文架構................................................ 4
第二章 背景知識與相關研究 .................................. 5
1 影像檢索技術.............................................. 5
1.1 基於文字影像檢索...................................... 5
1.2 內涵式影像檢索........................................ 6
1.3 影像檢索方式.......................................... 6
1.4 文字檢索與內涵式影像檢索之優缺點比較.................. 9
1.5 檢索介面............................................. 10
2 相關性回饋技術........................................... 10
3 其它檢索技術............................................. 11
4 雲端運算內容與服務之相關技術............................. 12
4.1 雲端平行處理技術..................................... 13
4.2 Hadoop............................................... 14
第三章 本論文之系統架構設計與影像特徵 ...................... 23
1 系統架構與功能概述....................................... 23
2 系統運算架構分析......................................... 25
3 MPEG-7 標準 .............................................. 29
4 影像特徵................................................. 31
4.1 色彩結構描述子....................................... 31
4.2 色彩可調性描述子..................................... 33
4.3 色彩佈局描述子....................................... 34
4.4 邊緣直方圖描述子..................................... 36
4.5 同質紋理描述子....................................... 39
5 相似度量測............................................... 40
6 正規化相關係數演算法..................................... 40
第四章 實驗模擬與結果 ..................................... 44
1 系統參數................................................. 44
2 檢索介面................................................. 47
3 實驗結果................................................. 51
3.1 節點的效能評估....................................... 51
3.2 檢索的準確性與分析................................... 55
第五章 結論與未來展望 ..................................... 62
1 結論..................................................... 62
2 未來研究方向............................................. 63
參考文獻 ..................................................... 64

                                

[1] Ling-Yu Duan, Feng Gao, Jie Chen, Jie Lin, Tie-Jun Huang, “Compact descriptors for
visual search,” IEEE MultiMedia, vol. 21 , Is.3 pp. 30-40, 2014
[2] D. B. Terry, “Caching hints in distributed system,” IEEE Trans. Software Engineering,
vol. se-13, no. 1, Jan. 1987.
[3] M. Batko, V. Dohnal, D. Novak, and J. Sedmidubsky, “MUFIN: A multi-feature
indexing network,” IEEE 2nd Int. Workshop on Similarity Search and Application, 2009.
[4] MUFIN, http://mufin.fi.muni.cz/imgsearch/, 25 October 2013.
[5] Overview of the MPEG-7 Standard (version 7), ISO/IEC JTC1/SC29/WG11 N4674,
Mar. 2002.
[6] Y. Rui, T. S. Huang, M. Ortrga, and S. Mehrotra, “Relevance feedback: a power tool in
interactive content-based image retrieval,” IEEE Trans. Circuits and Systems for Video
Technology, vol. 8, pp. 644-655, 1998.
[7] J.-J. Chen and C.-Y. Liu, “A universal query mechanism for similarity retrieval based on
shape information in image database,” IEEE International Conference on Acoustics,
Speech, and Signal Process, vol. 4, pp. 3676-3679, 2002.
[8] J.-J. Chen, C.-Y. Liul, Y.-S. Hunang, and J.-W. Hsieh, “Similarity retrieval in image
databases by boosted common shape features among query images,” IEEE The 2nd
Pacific-Rim Conference on Multimedia, PCM 2001, pp. 285-292, Oct. 2001.
[9] The Apache Hadoop. http://hadoop.apache.org/
[10] Y. Rui, T. S. Huang, and S.-F. Chang, “Image retrieval: current techniques, promising
directions, and open issues,” IEEE Journal of Visual Communication and Image
Representation, vol. 10, pp. 39-62, 1999.
[11] Y. Rui, T. S. Huang, M. Ortrga, and S. Mehrotra, “Relevance feedback: a power tool in
interactive content-based image retrieval,” IEEE Trans. Circuits and Systems for Video
Technology, vol. 8, pp. 644-655, 1998.
[12] C. Carson, S. Belongie, H. Greenspan, and J. Malik, “Blobword: image segmentation
65
using expectation-maximization and its application to image query,” IEEE Trans. PAMI,
vol. 24, no. 8, Aug. 2002.
[13] J. R. Smith, S.-F. Chang, “An image and video search engine for the world-wide web,”
In IS&T/SPIE Symposium on Electronic Imaging: Science and Technology (EI'97) -
Storage and Retrieval for Image and Video Databases V, San Jose, CA, February 1997.
[14] J. R. Smith, “Integrated spatial and feature image system: retrieval, compression and
analysis,” PhD thesis, Graduate School of Art and Sciences, Columbia University, Feb.
[15] A. Pentlandm, R.W. Picard, and S. Sclaroff, “Photobook: content-based manipulation of
image databases,” IEEE Journal of Computer Vision, vol. 18, pp. 233-254, 1996.
[16] K.-S. Candan et al, “Scalable multimedia data processing in server clusters,”
IEEE MultiMedia, pp.3-5, 2010.
[17] A. Verma et al, “Scaling genetic algorithms using MapReduce,” Int. Conf.
Intelligent Systems Design &Applications, 2009.
[18] W. Xu et al, “Detecting large-scale system problems by mining console logs,” in
Proc. of the ACM SOSP, 2009.
[19] S. Ghemawat, H. Gobioff, and S.-T. Leung, “The Google File System,” in ACM
Symp. on SOSP, 2003.
[20] J. Dean and S. Ghemawat, “MapReduce: simplified data processing on large
clusters,” Comm. of the ACM, vol. 51, pp.107-113, 2008.
[21] Apache Lucene. http://lucene.apache.org/ , 22 November 2013.
[22] Corel Image, http://www.brothersoft.com/downloads/corel-image-database-dvd.html,
18 July 2014.
[23] Caltech 101, http://www.vision.caltech.edu/Image_Datasets/Caltech101/, 7 July 2014.
[24] Flickr, http://www.flickr.com/groups/tw/, 6 March 2014.
[25] CDVS_dataset, http://wg11.sc29.org/visual-search/, 23 January 2014.
[26] MPEG-7,http://sabia.tic.udc.es/gc/Contenidos%20adicionales/trabajos/Imagenyvideo/vi
deo%20digital/futuro.html,12 June 2014.

簡易檢索 / 詳目顯示

相關論文