| 研究生: |
林佳駒 Lin, Chia-Chu |
|---|---|
| 論文名稱: |
以特徵為基礎的模糊演算法進行多波段-多極化NASA/JPL POLSAR影像分類之研究 Multi-channel and Multi-polarization NASA/JPL POLSAR Image Classification by a Features-Based Fuzzy Approach |
| 指導教授: |
蔡展榮
Tsay, Jaan-Rong |
| 學位類別: |
碩士 Master |
| 系所名稱: |
工學院 - 測量工程學系 Department of Surveying Engineering |
| 論文出版年: | 2003 |
| 畢業學年度: | 91 |
| 語文別: | 中文 |
| 論文頁數: | 109 |
| 中文關鍵詞: | 異質性 、物件導向 、影像分塊 |
| 外文關鍵詞: | heterogeneity, object-oriented, image segmentation |
| 相關次數: | 點閱:89 下載:1 |
| 分享至: |
| 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 |
合成孔徑雷達日夜皆可施測,且較不受天候限制,提供了相當多樣的遙測資訊,而歐洲太空總署(ESA)和加拿大太空總署(CSA)分別於2001年和2002年各發射了一顆多極化合成孔徑雷達遙測衛星,未來將提供比單極化合成孔徑雷達更豐富的地表雷達散射資訊,另外,由於受到解析力的限制,在一個像元內常會混雜著多種的地表覆蓋物,隨著各類感測器技術的提升,解析力變細,此類的情形也將逐漸的降低,而能獲取較細微的地物類別。
由於多波段多極化能提供比單波段單極化更多的地表之雷達散射資訊,因此本研究使用多波段多極化來進行雷達影像之分類,進而提升分類的精度。而在像元內常混雜著其它地物類別之雷達散射資訊,故在本文中將模糊理論引入影像分類中,使其分類時能容許類別混雜之情形存在,此外,相較於以像元為基礎的影像分類法而言,以區塊為基礎的影像分類法能減少分類處理的資料量,且能容許摻雜少許不同的地表物(包含雜訊),以增加資料的可利用性。在實驗中使用eCognition軟體來進行各類處理,由於在eCognition所使用的參數需由使用者自行輸入,在本文中將設計一方法來給定其參數值,以提升eCognition之自動化程度,另外,除了區塊特徵外,本文提出更多的特徵來提升分類精度,包括內部特徵、拓樸特徵與脈絡特徵。
實驗成果顯示使用最鄰近分類法其整體精度為91.31%,kappa coefficient為86.44%,經由吾人利用物件導向影像分類來提升分類精度,其整體精度為98.27%,kappa coefficient為97.23%,此處之物件導向分類法不單只使用以區塊為特徵來進行影像分類,吾人加入了物件導向之特繼承、類別關係於影像分類內,由實驗可發現,此分類法可分離位於特徵空間內較相近之類別。
Synthetic Aperture Radar (SAR) has less dependence on climate condition and can be operated day and night. It provides rich information for remote sensing. The ESA and CSA launched a satellite carried with a multi-polarization SAR system in 2001 and 2002, respectively. Such multi-polarization SAR systems will provide more information than single-polarization SAR system. Due to a limited resolution, a pixel often contains various kinds of land covers. Nevertheless, current advanced sensor technology makes a finer SAR resolution available. Thus, a finer object can be distinguished using such finer SAR data.
Because multi-channel and multi-polarization SAR data provide more information about the scattering characteristics of the earth’s surface, they enable a more accurate classification than single-channel and single-polarization SAR do. Hence, multi-channel and multi-polarization SAR data are adopted in this thesis to study the classification accuracy. Moreover, the fuzzy approach is utilized to take mixed pixels and regions with different classifications into account. Besides, region-based image classification provides a more efficient classification than pixel-based approach, since it reduces the data volume needed in classification operation and tolerates a mixed segment. It exploits as much available information as possible for image classification. Both region-based and pixel-based classification will be studied, too. Furthermore, this thesis proposes a approach defining a suitable threshold used in the eCognition software package. It enables a more automatic image classification using the eCognition system. To the other hand, this thesis defines more features, such as intrinsic features, topological features, and context features, to be applied and verifies that they can increase the accuracy of SAR image classification.
Test results show that the overall accuracy and kappa coefficient of object-oriented image classification is 98.27% and 97.23%, respectively, which is better than the nearest-neighbor approach (91.31% and 86.44%). The object-oriented classification approach not only uses region-features, but also some specific relationships such as inheritance and class. They also verify that a shorter separability distance in the adopted feature space is available in this classification method.
1. 王志添、陳家堂、張芮榛、梅素華、陳錕山,「利用空載全偏極SAR資料模擬ENVISAT-1及RADARSAT-2衛星SAR影像於地表分類之應用」,第十九屆測量學術及應用研討會論文集,p497~p501,1998。
2. 江凱偉,「小波分解輔助影像紋理區之分塊」,國立成功大學測量工程研究所碩士論文,台南,1997。
3. 卓榮邦、楊士毅和張金全等,「現代雷達原理」,電子工業出版,北京,1991。
4. 陳卉瑄,「差分合成孔徑雷達應用於偵測集集地震地形變之研究」,國立成功大學地球科學研究所碩士論文,台南,2001。
5. 陳鴻緒,「使用ERS資料與SAR干涉技術在台灣地區求定DEM之實務探討」,國立成功大學測量工程研究所碩士論文,台南,2001。
6. 陳家堂、陳錕山,「全偏極合成孔徑雷達於地表分類之研究」,第二十一屆測量學術及應用研討會論文集,p419~p426,2002。
7. 莊雲翰、陳繼藩,「知識庫影像判釋系統-以IKONOS衛星影像為例」,第二十一屆測量學術及應用研討會論文集,p179~p186,2002。
8. 馮德益、樓世博,「模糊數學:方法與應用」,科技圖書股份有限公司,台灣,1988。
9. 蔡忠孝,「多尺度彩色影像分塊之研究」,國立成功大學測量工程研究所碩士論文,台南,2000。
10. 劉治中、蕭國鑫和李元炎,「遙測與資訊結合應用於土地利用分類探討~以水稻田判釋為例」,第十八屆測量學術及應用研討會論文集,pp.673~682,1999。
11. 藎壚,「實用模糊數學」,亞東出版發行,臺北市,1991。
12. A. Freeman,“SAR Calibration: An Overview”,IEEE Trans. on Geoscience and Remote Sensing,Vol. 29,pp. 1107 –1121,1991。
13. A. P. Witkin,“Scale Space Filtering:A New Approach to Multi-scale Description”,Proc. of ICASSP,San Diego,CA March, pp. 39A.1.139A.1.4,1984。
14. Anhua Chu,“AIRSAR Integrated Processor Documentation:DATA FORMATS”, Version 0.16 December 16,2002。
15. C. Bouman and B. Liu,“Multiple Resolution Segmentation of Textured Images”,IEEE Transactions on Pattern Analysis and
Machine Intelligence,Vol. 13,No. 2,pp. 99-113,1991。
16. DEFINIENS AG,eCoginition user’s guide,2001。
17. Dubois, P. C. and L. Norikane,“Data volume reduction for imaging radar polarimetry”,IGARSS '87 (Ann Arbor, MI), in Proc,pp. 691-696,1987。
18. E. Salari and Z. Ling,“Texture Segmentation using hierarchical Wavelet Decomposition”,Pattern Recognition,Vol. 28,No. 12, pp. 1819-1824,1995。
19. Floyd M. Henderson and Anthony J. Lewis,“Principles & Application of IMAGING RADAR”, J. Wiley,New York,1998。
20. GlobeSAR-2 Program,“Educational Resources for Radar Remote Sensing”, Canada Centre for Remote Sensing,Canada,2001。
21. Gunter Schreier,“SAR Geocoding: Data and Systems”,Wichmann,Karlsruhe,1993。
22. IVITS, E. & B. KOCH,“Object-Oriented Remote Sensing Tools for Biodiversity Assessment: a European Approach”,In Proceedings of the 22nd EARSeL Symposium,Prague, Czech Republic, Millpress Science Publishers,Rotterdam,Netherlands,2002。
23. J.A. Richards,“Remote Sensing Digital Image Analysis”, Springer-Verlag,Berlin,1994。
24. J. J. Van Zyl and F. T. Ulaby ,“Scattering matrix representation for simple targets”,Radar Polarimetry for Geoscience Applications,Artech House,1990。
25. J. T. Tou and R. C. Gonzalez,“Pattern Recognition Principles”, Addison-Wesley Publishing Company, Reading, Massachusetts,1974。
26. L. A. Zadeh,“Fuzzy sets” ,Information and Control,vol. 8,pp.338-353,1965。
27. Leland E. Pierce, Fawwaz T. Ulaby, Kamal Sarabandi, and M. Craig Dobson, “Knowledge-Based Classification of Polarimetric SAR Images”,IEEE Trans. on Geoscience and Remote Sensing,VOL 32,NO. 5,1994。
28. Lillesand T. M. and R. W. Kiefer,“Remote Sensing and Image Interpretation”, John Wiley & Sons,New York,2000。
29. Lopes A., E. Nezry, R. Touzi, and H. Laur,“Structure Detection and Statistical Adaptive Speckle Filtering in SAR Images”,International Journal of Remote Sensing,Vol. 14,No. 9,pp.1735-1758,1993。
30. M. Baatz and A. Schäpe,“Multiresolution segmentation - An optimization approach for high quality multi-scale image segmentation”,in Angewandte Geographische Informationsverarbeitung,XII. Beiträge zum AGIT Symposium (Strobel et al. Hrsg.). Salzburg, pp 12-23,2000。
31. M. Helmut,“Abstraction and Scale-Space Events in Image Understanding”,International Archives of Photogrammetry and Remote Sensing,Vol. XXXI,Part B3,pp.523-528,1996。
32. NASA,instrument panel report。SAR:Synthetic Aperture Radar(Earth Observing System,Vol. IIf),Earth Science and Applications Division,NASA Headquarters,Washington, D.C.233p.,1989。
33. NASA JPL,Technical Manuals:AIRSAR General Reference Manual,2003。http://airsar.jpl.nasa.gov/documents/instrument.htm
34. P.B.G. Dammert, J.I.H. Askne and S. Kühlmann,“Unsupervised Segmentation of Multitemporal Interferometric SAR Images”,IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING,VOL. 37, NO. 5,pp. 2259-2271,1999。
35. Rafael C. Gonzalez and Richard E. Woods, “Digital Image Processing”, Addison-Wesley Publishing Company,Reading, Mass.,1992。
36. Robert A. Schowengerdt,“Remote sensing, models, and methods for image processing” ,San Diego,Academic Press,1997。