- 期刊
特徵匹配於衛星影像套合之應用
Satellite Image Registration Using Feature-Based Matching
摘要
為提升衛星遙測科技之應用，多重遙測影像間整合之需求日殷。因此數位影像間幾何位置套疊關係之建立，於應用上即成為不可或缺的資料前處理步驟。本研究將特徵匹配之機制應用於影像套合，冀能提升其自動化之完整性。以影像中之均調區邊界為特徵，利用一形狀矩陣作為描述元，將區塊的形狀特徵網路化並進行特徵匹配，藉由一對應篩選法則濾除特徵匹配中錯誤之對應。再由這些正確的對應區塊，以其重心位置為套合控制點，進行起始轉換係數求解，可找出並建立套合影像與參考影像間起始的相對坐標轉換關係。以此為出發點，可提供進一步套合控制點加密時，匹配位置之預估資訊，完成精密影像套合。本研究之主體分為特徵匹配並建立起始對應關係與精密影像套合兩部份；在特徵匹配並建立起始對應關係部份，主要程序如下：(1)利用形狀矩陣分別在套合影像與基準影像上對萃取出之影像區塊進行特徵描述，並進行相似性評估，產生對應候選；(2)以區塊之重心連線的位向與幾何關係，濾除錯誤的對應候選，建立兩影像之起始對應關係。在精密影像套合部份，萃取參考影像上的特徵點，利用前面所得到之對應關係，提供套合影像上匹配之預估位置，進行區域匹配。再依所得匹配之點對結三角網，進行影像之局部套合。
並列摘要
Data integration for multiple images is getting more important as more resource satellites become operational. Image registration is an indispensable preprocessing in integrating multi-source images. The automated densitification technique for registration control points is getting mature. However, the establishment of initial correspondence still needs efforts to increase the degree of automation. The objective of this investigation is to correspond the conjugate feature pattern those are extracted from a reference images and its counterpart, i.e., a sensed image. The major components of the proposed scheme include: (1) quantatively describing the feature patches using shape matrix, (2) similarity assessment between patch pairs, (3) considering the topologocal and geometrical relationship between candidate patch controids to filter out unreliable matching, (4) establishing initial transformation according to the reliable centroid pairs, and (5) applying fine matching to generate registration control point followed by a local transformation procedure.