四元素分解模型於全偏極雷達散射特徵應用與分析;Applications and analysis of full polarimetric SAR scattering characteristic based on four component decomposition model

NCU Institutional Repository > 研究中心 > 遙測科技碩士學位學程 > 博碩士論文 > Item 987654321/53582

jsp.display-item.identifier=請使用永久網址來引用或連結此文件: http://ir.lib.ncu.edu.tw/handle/987654321/53582

题名:	四元素分解模型於全偏極雷達散射特徵應用與分析;Applications and analysis of full polarimetric SAR scattering characteristic based on four component decomposition model
作者:	程姿諭;Cheng,Tzu-yu
贡献者:	遙測科技碩士學位學程
关键词:	全偏極;四元素分解模型;合成孔徑雷達;POLSAR) Synthetic;POLSAR) Synthetic Aperture Radar;POLSAR)Synthetic Aperture Radar;Polarimetric(Polarimetric (Polarimetric(Polarime;polarimetric;Four Component Decomposition;POLSAR)Synthetic Aperture Radar;POLSAR) Synthetic Aperture Radar;POLSAR)Synthetic Aperture Radar
日期:	2012-08-30
上传时间:	2012-09-11 17:59:46 (UTC+8)
出版者:	國立中央大學
摘要:	全偏極合成孔徑雷達( Fully Polarimetric Synthetic Aperture Radar, POLSAR)具全偏極特性，應用發展不同於單一波段合成孔徑雷達系統。其特色於相同目標物採用不同極化方式發射接收，因目標物結構特徵差異具有不同散射特徵。全偏極合成孔徑雷達特徵包含散射矩陣(Scattering Matrix)中除了線性偏極特徵外，尚包含偏極間交互作用項。以全偏極雷達特徵作為分類資訊相較於已往以部分偏極特徵為分類資訊，其在優勢於保有更完整之地物散射特性。因此利用全偏極合成孔徑雷達影像進行地表觀測，有利於完整的目標分析。本文討論四元素分解模型(Four-Component Decomposition Model)於全偏極影像資料並討論應用於不同目標物之影像結果。四元素分解模型為共變異矩陣(covariance matrix)或同調矩陣(coherence matrix.)中所分解出表面散射(single bounce scattering)、二次散射(double bounce scattering)、體散射(volume scattering)與螺旋散射(helix scattering)所組成。四元素分解模型應用於全偏極合成孔徑雷達影像有兩個問題需提出討論。一是要消除在建物區雷達照射方位的影響，同調矩陣分解前必須先修正方位。方位角經由最小化交叉極化項或從圓極化基底的相關係數相位推得。其中假設隨機偶極(randomly dipoles)的體散射模型，十分難以物理散射介質佐證。與其模型定義與資料並不匹配。二是在分解模型過程中會產生負功率問題。旋轉方位角後，降低在表面散射與二次散射產生負功率的狀況，但無法完全修正。本研究致力於應用此模型基於分解後的散射組成空間特徵向量變遷偵測。此組成由實際量測到的二階散射矩陣經由四元素分解模型經由統計特徵而獲得。並以分解模型為基礎，探討提升全偏極影像在地物散射特徵應用與變遷偵測能力。測試方面，資料為鰲鼓六幅不同時期RADARSAT-2全偏極合成孔徑雷達資料。針對稻作與沿岸沙洲進行四元素分解模型的應用與分析。透過全偏極合成孔徑雷達影像，應用於不同地表特性的多時序觀測，針對特定目標物(稻作、沙洲)個別討論。稻作部分，透過四元素法分解模型計算各個頻段背向散射係數值。配合當日實際地面調查拍攝稻作生長資料。影像中顯示，稻作依據不同生長週期具不同變遷特徵且符合稻作生長週期所表現出特性。藉此可利用衛星影像監測稻作生長情形。沙洲部分，由於西海岸線潮位變化幅度大，不同海面高度觀測到沙洲及沿岸蚵架所呈現的反應也不同。低潮位時，蚵架露出海面與雷達飛行方向正交，呈現二次反射的特徵。另外也可藉由體散射項觀測浪前緣的變化。高潮位時，沙洲幾乎被淹沒，海浪遇上較高地形而產生較大的擾動，使得海水淹沒的沙洲呈現體散射特徵。配合現地調查資料，四元素分解模型影像與現地資料調查結果相符合一致。較以往傳統分解方法，更具正確分辨率。未來應用於農業植被生長觀測或國土海岸線的監測都十分地有效益。A four‐component scattering model, being identified by single bounce, double bounce, volume, and helix scattering power contributions, was proposed by Yamaguchi to decompose fully polarimetric synthetic aperture radar (SAR) images based on covariance matrix and coherence matrix- the later bears better quantitative interpretation of SAR data and easier computation. To remove the effects of orientation in urban buildings with radar illumination, the coherence matrix is de‐orientated before decomposition. The orientation angle may be derived by minimizing the cross‐polarized term or derived from the phase of the correlation coefficient in the circular polarization basis. Among by Lee propose volume scattering is modeled with randomly oriented dipoles, it is difficult to prove if these basic scattering models fit the physical scattering media. The models definitely fail to match the data. An attempt has been made to incorporate the X-Bragg model modify the scattering model. The proposed algorithm will significantly reduce the number of negative pixels.This study is devoted to applying these model based decomposed scattering components as spatial feature vector to change detection. These components can be viewed as second order statistics like texture measures from the fact that these four components are derived from second‐order statistics of scattering matrix. Fully polarimetric SAR data from series Radarsat‐2 were acquired over western Taiwan coast. Change detection was then carried out by a simple threshold of the cumulative histogram of the difference feature vector according to a predefined constant false alarm rate (CFAR) value. Results indicate that the excruciating mis‐registration error was capably taken into account and removed without affecting the excellent detection performance.
显示于类别:	[遙測科技碩士學位學程] 博碩士論文

文件中的档案:

档案	描述	大小	格式	浏览次数
index.html		0Kb	HTML	1094	检视/开启

在NCUIR中所有的数据项都受到原著作权保护.

社群 sharing

数据加载中.....