工學院: 土木工程學研究所指導教授: 韓仁毓黃迺絜Huang, Nai-JieNai-JieHuang2017-03-132018-07-092017-03-132018-07-092016http://ntur.lib.ntu.edu.tw//handle/246246/278005槽溝挖掘方法為古地震學主要研究方法之一，藉由挖掘後的現地資料及槽溝剖面地層圖做為地震潛勢災害分析之依據，由於目前槽溝剖面地層圖之製作方式主要仰賴大量人力和簡易的照相記錄，不僅耗時費力，且成果品質不易掌握。因此，本研究目標為將發展以搭載著攝影機之地面光達系統產製槽溝剖面地層圖之流程，由現地取得槽溝三維坐標點雲後，將各測站點雲套合投影至最適投影平面，並以彩色點雲資訊為基礎，將產製之槽溝影像結合紋理特徵萃取進行影像分類。實驗成果顯示，所提出的方法可以由光達掃瞄之離散點雲產製槽溝影像，並能額外提供影像和光達點雲資料之尺度誤差量做為變形評估之依據，而在槽溝影像分類成果，依據過去的槽溝地層分層圖進行監督式分類，亦可有效地分出地層類別，分類整體精度為81.18%，Kappa值為0.7646。同時，在加入紋理分析後能與光譜分類結果有效結合，產製出含有紋理資訊之槽溝地層分層圖，以輔助地質學家後續進行歷史地震事件分析及相關應用。Excavating fault trenches is one of main method to study paleoseismology. However, the current approach which costs considerable time and manpower only relies on geologists to recognize the multiple faulting events. Therefore, this study constructs an automatic procedure to map geological sections by using terrestrial LiDAR systems integrated with cameras. The experiment results indicated that the proposed approach can provide high-quality images of fault trenches and amount of image deformation can also be simultaneously evaluated. In the supervised classification procedures, the classification accuracies are also encouraging (overall accuracy 81.18% and kappa 0.7646). Furthermore, by using an existing geological section produced in 2002 with the texture analysis and classification from the image, the major categories can be correctly identified. It gives solid evidence that the terrestrial LiDAR techniques can be efficiently extended to the applications in geological studies.8892128 bytesapplication/pdf論文公開時間: 2021/8/30論文使用權限: 同意有償授權(權利金給回饋本人)地面光達槽溝挖掘影像分析影像分類紋理分析Terrestrial LiDAR SystemTrench ExcavationImage AnalysisImage ClassificationTexture Analysisthesis10.6342/NTU201602647http://ntur.lib.ntu.edu.tw/bitstream/246246/278005/1/ntu-105-R03521118-1.pdf