高解析遙測影像之幾何改正、套合及特徵萃取(I); Geometric Correction, Registration and Feature Extraction for High Resolution Remotely Sensed Images (I)

NCUIR > Research Center > Center for Space and Remote Sensing Research > Research Project > Item 987654321/40053

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/40053

Title:	高解析遙測影像之幾何改正、套合及特徵萃取(I);Geometric Correction, Registration and Feature Extraction for High Resolution Remotely Sensed Images (I)
Authors:	陳良健;劉說安
Contributors:	太空及遙測研究中心
Keywords:	土木水利工程類
Date:	2001-12-01
Issue Date:	2010-11-03 11:42:25 (UTC+8)
Publisher:	行政院農業委員會
Abstract:	本三年期計畫主要針對高解析遙測影像進行幾何處理及特徵萃取.幾何處理包括幾何改正及影像間之套合.考量之資料包括高解析衛星影像, 空載多譜掃描儀影像, 及數化空照像片.三年為期之工作劃分如下, 第一年( 即本年度 )將以高解析衛星影像之幾何處理為主, 包括方位重建以及嚴密之正射化方法開發.第二年, 著重於上列三種影像間之套合, 並由所建之機制完成地面控制點之量測.第三年則進行特徵物( 如房屋、道路 )之萃取.考慮時程, 將以半自動之方式針對房屋部份進行處理.第一年之工作為建立幾何改正之模式包括方位重建及正射化程序.在方位重建部份需要考慮姿態參考坐標系, 局部軌道坐標系, 地心坐標系, 地心赤道坐標系, 地理坐標系, 及投影坐標系相互間之正逆轉換.此外, 需利用軌道載體參數進行概略方位參數之構建, 再依據少量的地面控制點依呈像時共線之物理條件進行方位參數之精化.方位精化過程中, 依控制點之物空間與像空間坐標進行最小二乘過濾, 以降低雜訊之干擾.方位重建後即進入正射化之程序.除了考慮DTM進行影像校正外, 亦將考慮房屋所造成數值表面模型( Digital Surface Model )有高度不連續的情形.此時遮蔽現象是考慮之重點.因此需建立一個新的運算法則, 以使校正模式更為完整.此一運算法則將由呈像中心與地表點所構建之空間向量進行判斷遮蔽的範圍, 進而達成具房屋涵蓋區之高解析影像正射化.在資料測試方面則將使用高解析衛星IKONOS影像.因為高解析衛星幾何處理仍需地面控制點.若此種控制點可由更高解析力之數化空照像片依套合技術自動化產生控制點, 則將可提升其處理之效能.此二種資料亦可由套合/匹配技術產生空載掃描儀影像之控制點, 則其多譜特性將更可發揮.這些套合技術之開發屬第二年之工作.本年度之預期效益包括: ( 1 )構建Pushbroom掃描儀影像之方位模組.不但可適用於本計畫, 同時, 將保留修改之彈性以適用其它類似之衛星.目標為高精密度及低控制點需求.( 2 )建立含人工建物之影像校正( 正射化 )程序.以期在地理資訊系統中可與其它空間資料整合. This three-year project is to investigate the geometric processing and feature extraction for high resolution remotely sensed images.Geometric processing includes geometric correction and image registration.Target images include satellite and airborne images.The major work in each year is stated as follows.In the first year, the major effort will be in the orientation modeling and orthorectification.The second year work will emphasize on the image registration among different image sources.The developments in the second year will be used to measure ground control points.In the third year, semi-automatic feature extraction, such as man-made buildings, will be performed.The central work in the first year includes orientation modeling and orthorectification.In the orientation modeling, coordinate transformation between attitude reference frame, local orbit system, geocentric coordinate, geocentric equatorial coordinate, geographic coordinate system, and map projection system need to reconstruct the approximate orientation parameters.Then ground control points will be used to refine the parameters.In which, least squares filtering will be considered to minimize the influence of noises.The next step is to do the orthorectification.In addition to the traditional procedure for orthorectification, surface discontinuity due to man-made buildings should be taken into account.Accordingly, a new algorithm considering digital surface model ( DSM )will be developed.In the treatment for surface discontinuity, detection of hidden areas and processing afterwards are the most important steps.IKONOS images will be used in the experiment.Although limited but still needed is the ground control information.The techniques of registering satellite images and aerial images may be extended to perform the GCP generation.The developments for registration technique will be considered in the second year.The proposed project may achieve the following goals.The first is to reconstruct orientation parameters for pushbroom images.The techniques may be extended to other satellite images with pushbroom devices.Low number of GCP requirements are expected.The second goal of this investigation is to establish a procedure for high resolution images where surface discontinuity is considered.Thus, the orthorectified imaged may be integrated with other spatial data in a geographic information system. 研究期間：9001 ~ 9012
Relation:	財團法人國家實驗研究院科技政策研究與資訊中心
Appears in Collections:	[Center for Space and Remote Sensing Research ] Research Project

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	436	View/Open

社群 sharing

Loading...