應用無人工佈標之近景攝影測量建立災害邊坡數值地形模型

Using Close-Range Photogrammetry to Reconstruct Digital Terrain Model of Post-Disaster Slope without Artificial Targets

鐘琬婷(Wan-Ting Chung)

指導教授 ： 張哲豪

國立臺北科技大學/工程學院/土木與防災研究所/碩士(2011年)

https://doi.org/10.6841/NTUT.2011.00039

摘要

近幾年空間技術發展已趨近完整。在雷射掃描技術、航空與地面攝影測量技術，已經可以短時間內獲取高精度之三維資訊。台灣長期受到地震、豪雨、颱風等影響，自然因素發生頻繁，所以常造成災害發生，當災害發生，邊坡、土石狀況極為不穩定，本研究以木柵貓空纜車站指南宮站後方之猴山岳做為研究區域，期望以快速且動用最少人力、物力及考量測量人員安全因素之測量方法，快速紀錄災後現場的三維空間資訊，並建立可供展示及量測之三維模型，即時反映災害現場情況，可提供日後的研究與分析。
本研究將介紹以無人工佈標之近景攝影測量方法，以全測站量測獲取邊坡上特徵點坐標，應用軟體完成災害邊坡模型重建。為了評估研究中近景攝影測量重建之邊坡模型精度，精度檢核分成兩個部分：單點量測精度與模型精度。以全測站測得之控制點坐標，從中挑選出5點由控制點改為檢核點不解算外方位，求出全測站與立體量測各方向之殘差；模型精度檢核，研究使用高精度地面光達量測(Ground LiDAR)生成之模型作為檢測標準，比較兩個模型之間的差異。實驗最後以虛擬實境模組語言（Virtual Reality Modeling Language；VRML)展示成果。

關鍵字

近景攝影測量 ； 地面光達 ； 三維模型 ； 特徵點 ； 邊坡 ； 滑動邊坡 ； 根面積比

並列摘要

Spatial technology has undergone rapid development in recent years. In laser scanning technology as well as aerial and terrestrial photogrammetry, highly accurate three-dimensional data can now be acquired in relatively short time. Taiwan is a place prone to calamities such as earthquakes, heavy rainfall and typhoons, which in turn result in highly unstable slope conditions and landslides. This study targets the area of Houshanyue behind Zhinan Temple Station of the Moakong Gondola in Muzha as a site of interest. Our purpose is to develop measurement methods which require minimal manpower and resources and in the meantime ensure personnel safety; this can facilitate instant documentation of three-dimensional data at disaster sites. Three-dimensional models constructed from such data can in turn be utilized for display and measurement purposes, providing immediate information from the sites of disaster and also material for subsequent study and analysis.
This study introduces a close-range photogrammetry method without artificial targets. We acquired feature point coordinates via total station measurement and performed software-assisted model reconstruction of slopes at disaster sites. To evaluate the accuracy of slope models, accuracy verification was focused on two aspects: single-point measurement accuracy and model accuracy. For the first aspect, among the control point coordinates obtained from the total station, five control points were selected and altered to checkpoints of non-solving exterior orientation, and the residuals on the model and between checkpoints were thus calculated. As for model accuracy verification, models generated via high accuracy ground LiDAR measurement served as standards for comparison of two different models. Final results were displayed in the form of Virtual Reality Modeling Language (VRML).

並列關鍵字

Close-Range Photogrammetry ； LiDAR ； 3D model ； Feature ； Slope ； landslides ； Root Area Ratio

參考文獻

[40] 石佑偉（2010），以物空間匹配方法重建三維顏面模型，碩士論文，國立台北科技大學土木與防災所。

[48] 孫浩淳（2010），採用根系纖維叢分析法的網路邊坡穩定程式開發，碩士論文，國立台北科技大學土木與防災所。

[47] 馬國宸（2008），土壤水分移動與植生根系對邊坡穩定之研究，博士論文，國立台灣大學生物環境系統工程學研究所。

[59] 蕭國鑫、劉進金、曾義星、王晉倫（2010），三維雷射掃描應用於崩塌土石量化之研究，航測及遙測學刊第十五卷第1期第97-109頁。

[51] 陳秀婷（2006），植生土壤剪力強度之實驗量測與數值分析，碩士論文，國立中央大學土木工程研究所。