傳統上，森林覆蓋製圖一般採用航照判釋配合現場調繪方式來進行，其往往需要投入大量的人力與物力，故利用遙測技術節省調查成本，已成為森林覆蓋製圖研究之重要議題。本研究以ADS40航攝數位影像為研究材料，並採用物件導向影像分析法（object-based image analysis, OBIA）進行森林覆蓋製圖。研究中使用OBIA處理影像資訊，並透過分類迴歸樹演算法（classification and regression tree, CART）整合各式影像分類資訊來源，同時利用階層式分類（hierarchical classification）概念進行影像分類，森林覆蓋型分類的總體精確度為77.60%，總體Kappa值為0.7196。其中非植生類別的分類成果較佳，而植生類別的分類精度則較低。高解析度影像分類須考慮十分繁複之因子與分類方法組合，並配合專家知識，方可得較佳成果。
Traditional forest cover type mapping is based on aerial photograph interpretation and field survey, it requires a lot of manpower and cost, and new forest inventory methods based on remote sensing that have been an important issue. In this study, ADS40 digital aerial photographs were used as materials, and we used object-based image analysis for forest cover type classification. We used object-based image analysis and classification and regression tree (CART) to class forest cover types. We used the concept of hierarchical classification for image classification. The overall accuracy of forest cover type classification is 85.75%, and the overall Kappa statistic is 0.72. The classification accuracy of non-vegetation classes is high, but for vegetation classes is slightly worse. High resolution image classification needed to consider very complex factors and combination of classification techniques, and also with expert knowledge that can get better results.
