本文研究影像紋理分析之理論，並針對空載多光譜影像中，不同地物像元在多維 度光譜空間分佈的特性，分析其各項影像紋理特徵，進而建立不同地物類之空載遙測影像紋 理特徵資料庫， 以為後續影像分類之用。 本研究選用臺中市地區十二個波段之 Daedalus DS-1260 空載多譜掃瞄影像進行分析，選取不同地物之子區域影像，以灰調空間相依機率矩 陣法計算各項紋理特徵值，結果顯示，影像紋理資訊可提供地物空間分佈特性之變化，作為 輔助遙測影像分類地物光譜辨識時之參考。 郇m及選手在安全及訓練上的資訊。並建議選 手在做此高難度動作時，在離槓前之大迴環底部附近應加上下半身鞭打動作，以增加離槓垂 直速度、空中高度及離槓後之旋轉速度; 注意轉體時雙臂收緊以利轉體。

　　　　　This paper examines the theory of texture analysis for airborne multispectral images. According to their distributions in the multi-dimensional spectral space, textural features of various types of land cover are analyzed such that the texture database of remotely sensed airborne images of land cover can be established for use in following image classification. The 12-channel Daedalus DS-1260 airborne multi-spectral images of Taichung city were used and sub-images of different kinds of land cover were chosen for texture discrimination analysis. The gray-tone spatial-dependence probability-distribution matrix algorithm was applied to extract texture information of the interested land covers. Empirical results show that image texture information reveals the variation of spatial distribution of land covers and provides valuable information to the spectral classification of remotely sensed images. t, they should perform a beat swing near the bottom of the giant to increase the CM vertical velocity flight height and the rotation speed after bar release. When twisting in flight, arms should be held tightly as much as possible.