English  |  正體中文  |  简体中文  |  Items with full text/Total items : 28611/40649
Visitors : 650758      Online Users : 44
RC Version 4.0 © Powered By DSPACE, MIT. Enhanced by NTU Library IR team.
Scope Adv. Search
LoginUploadHelpAboutAdminister

Please use this identifier to cite or link to this item: http://ntour.ntou.edu.tw:8080/ir/handle/987654321/33656

Title: 遙測衛星影像資料融合及平行處理技術應用於台灣海岸線變遷研究
Authors: 張陽郎;梁文耀;張麗娜;方志鵬
Contributors: 國立臺灣海洋大學:通訊與導航工程學系
Keywords: 資料融合;平行處理;海岸線變遷;地質遙測學;平行模擬退火特徵模組選取;評定模型;動態分散式平行計算環境即時嵌入式
coastline change detection;parallel positive Boolean function;parallel simulatedannealing band selection;Logit model predictor;real-time embedded system
Date: 2008-08
Issue Date: 2013-05-08T06:29:24Z
Publisher: 行政院國家科學委員會
Abstract: 摘要:由於人為大量的污染破壞,導致「溫室效應」(the greenhouse effect) 持續擴大,逐年改變全世界的氣候,並將造成全球部分島嶼的消失。臺灣四面環海,以海洋立國,海岸線環境變化原因錯綜複雜,不斷地影響到「臺灣海岸線」的快速變遷,如何運用有系統的「災害管理」及「地質遙測」技術,來有效管理海岸線的變遷,將是我們急需克服的課題。本研究計畫提出一個衛星遙測影像「資料融合及平行處理」的新方法,用以研究台灣地區「海岸線變遷」及「災害管理」的問題,並藉由先前研究所累積之寶貴經驗,尋找出一個適合應用於「高維地質遙測資料融合」的最佳多核心多處理器「平行布林函數」分類方法,同時提出一個新改良「平行布林函數」的監督式分類方法,稱為平行k-way半擬陣 (Parallel k-way Semi-Matroid, PKSM) 分類器。PKSM是以k-way tree為架構,每一個節點均由正、負樣本的集合所組成的,利用正、負樣本可以計算出每類別所佔之百分比,亦即是特定的類別與其他類別的比。PKSM是使用一個不平衡的k路樹狀結構,其中每個葉節點(leaf node) 均符合半擬陣 (Semi-Matroid)的架構,其代表了不同大小的區塊及所有類別的分佈。k路樹是建構在某個任一特定的類別上,根據不同類別間的統計機率,來判定是否停止切割產生新的子空間,以及哪個類別因歸屬於哪個子空間。PKSM學習模型透過不同類別的正、負樣本,解決了「平行布林函數」分類器,正、負樣本個數需要平衡的問題,就分類的正確率而言,PKSM優於傳統的分類器,據此進而提升衛星遙測影像「海岸線變遷」分類辨識及變遷預測率。希望藉由嚴謹的理論探討及深入的推導分析,以實作的結果,驗證所提之新架構優於傳統的地質遙測分類及變遷預測方法。尤其近年來因「溫室效應」的影響,台灣飽受氣候異常、颱風、地震等天然災害的肆虐,造成海岸線的快速變遷,如何利用衛星遙測影像「資料融合及平行處理」的方法,用以有效控制天然災害及管理土地資源,並藉由整合「地球科學」、「地質科學」、「遙測科技」、「資訊科學」及「電機工程」等不同學科研究,來跨領域探討台灣地區特有的「海岸線變遷」與「災害管理」等相關的議題。
abstract:A novel study is proposed for automatic coastline change detection using parallel computing and data fusion techniques with geological remote sensing images. The planning method is based on fusion of high-dimensional remote sensing images of the same scene collected from multiple sources. It will present a framework for fusion of multisource remote sensing images, which consists of three algorithms, referred to as parallel positive Boolean function (PPBF), parallel simulated annealing (PSA) band selection and Logit model predictor (LMP). Based on our previous proposed complete modular eigenspace (CME) method which was designed to extract the simplest and most efficient feature modules, the PPBF method is intended to improve the performance of the extracted CME features optimally by modifying the original positive Boolean function classifier operated in parallel. The PSA method is designed to extract features by a new defined multi-dimensional modular eigenspace (MME) and further to optimize the modular eigenspace, while the MME is performed to predict the transition of land-cover changes from different data sources. The performance of the proposed method, implemented by a new defined dynamic distributed mobile computing environment (D2MCE) in a real-time embedded system, is evaluated by fusing MODIS/ASTER (MASTER) hyperspectral and the synthetic aperture radar (SAR) images. It’ll be expected that the experiments will demonstrate the availability of the proposed PSA/LMP/PPBF approaches. It is an effective method for both the feature extraction of the high- dimensional fused data and information mining. It can also improve the classification accuracies and prediction rates significantly compared to the conventional geological remote sensing classifications and coastline changes detections.
Relation: NSC97-2116-M027-002
URI: http://ntour.ntou.edu.tw/handle/987654321/33656
Appears in Collections:[通訊與導航工程學系] 研究計畫

Files in This Item:

File Description SizeFormat
index.html0KbHTML168View/Open


All items in NTOUR are protected by copyright, with all rights reserved.

 


著作權政策宣告: 本網站之內容為國立臺灣海洋大學所收錄之機構典藏,無償提供學術研究與公眾教育等公益性使用,請合理使用本網站之內容,以尊重著作權人之權益。
網站維護: 海大圖資處 圖書系統組
DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU Library IR team Copyright ©   - Feedback