以遙測影像的多目標決策之崩塌地影像判釋：自組織映射圖與連續型粗集合離散化混合模型之研究

The Landslide RS Image Classification through Hybrid Model of SOM and Discrete Rough Sets

萬絢(Shiuan Wan) ； 劉宜麟(Yi-Lin Liu)

《水保技術》 8卷1期 (2013/01) Pp. 16-24

摘要

水庫附近的環境往往影響水庫水品質和淤積的狀況，然而水庫多是大範圍的地區非仰賴衛星影像不可。本研究使用衛星影像原始光譜與DEM的高程等相關資料，並計算坡度高程與植生指標作為影像整合多類別分類的資訊，再用（a）自組織映射圖（Self-Organizing Map; SOM）及（b）粗糙集理論（Discrete Rough Set; DRS），作為影像空間特徵分類方法，以達快速判釋崩塌地，有效降低時間及人力之目的。實驗區在萬大水庫，選擇了30筆崩塌地樣本的光譜值，第一步先就高程和坡度等資訊，建立門檻值，再以上述（a）、（b）兩種方法，以光譜值與八個植生因子分別進一步判釋崩塌地與非崩塌地。我們設計了一個以光譜配合坡地門檻值配合光譜資訊的改善崩塌地判釋方法，進行一系列的分析，產生多目標的分類方法，對未來坡地災害的管理有一定貢獻。

關鍵字

自組織映射圖 ； 粗糙集理論 ； 植生指數 ； 坡地門檻值

並列摘要

The reservoir environment often affects the water quality of the water of reservoir and the deposition conditions. However, the reservoir is more than a wide range of areas which must be relied on satellite image data. This paper presents a multi-category image classification. The slope of DEM and normalized difference vegetation index and other modification of spectrum information are used. The (a) self-organizing map and (b) Discrete Rough Set, are then used to achieve rapid interpretation are landslides and reduce the time consuming. We selected 30 samples of landslide spectral values surrounding the Wan-Da reservoir. Following (a) and (b), the classifier is used to differentiate landslide and non-landslide area. The results showed that spatial characteristics of the two image classification methods. We also examined 8 different indicators of vegetation. We design a novel method on how to effectively improve the performance of classification through threshold method. Regardless of the slope thresholds, the accuracy can only reach 78 %. However, if considering the thresholds, the accuracy can be improved about 10%.

並列關鍵字

Self-Organizing Map ； Discrete Rough Set ； Vegetation Index ； Slope Thresholds