- 學位論文
以半擬陣為基礎架構之k-way分類器應用於高光譜遙測影像
A Novel k-way Tree Classifier Based on Semi-Matroid Structure Applied to Hyperspectral Remote Sensing Images
摘要
在本篇論文中，我們提出一個新的監督式分類方法，為了監督多光譜遙測影像，稱為k-way Semi-Matroid (KWSM) 分類器。KWSM是以k-way tree為架構，是一個有組織系統的樹，每一個節點均由正、負樣本的集合所組成的，利用正、負樣本我們可以計算出每類別所佔之百分比。換而言之，所佔之百分比是特定的類別與其他類別的比。KWSM是使用一個不平衡的k路樹狀結構，其中每個葉節點(leaf node) 均符合半擬陣的架構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和哪一個類別歸屬於該子空間。KWSM學習模型透過不同類別的正、負樣本，就分類的正確率而言，KWSM優於傳統的分類器。為了計算我們所提出的KWSM效能，我們是利用Pacrim II的計畫MODIS/ASTER airborne simulator (MASTER)和airborne synthetic aperture radar (AIRSAR)的影像作為地表分類的依據。由實驗結果可知，KWSM的結果是我們所期待的。
並列摘要
In this thesis we present a new supervised classification method, referred to as the k-way semi-matroid (KWSM) classifier, for supervised classification of multisource remote sensing images. The proposed KWSM is organized by a k-way tree in which every node (semi-independence part in semi-matroid structure) is composed of a set of D-dimensional positive and negative labeled samples as represented as a percentage, i.e. the corresponding ratio of number of samples between a specific class and other classes. KWSM uses an unbalanced k-way tree in which each leaf node with semi-matroid structure has different region sizes regarding the distributions of all classes. By constructing the k-way tree based on each specific class, the statistical ratios between different classes are then compared as a basis for stopping the new subspace separating and identifying which class belongs to which subspace. By delivering both positive and negative samples of different classes to KWSM learning modules, KWSM outperforms traditional classifiers in terms of classification accuracies. The effectiveness of the proposed KWSM is evaluated by fusing MODIS/ASTER airborne simulator (MASTER) images and airborne synthetic aperture radar (AIRSAR) images for land cover classification during the Pacrim II campaign. According to experimental results, the KWSM performs as we expect.