本研究利用無人飛行載具空中拍攝之垂直正拍影像以及衛星定位儀(GPS)地面控制點的佈設做為建圖材料，透過商用套裝軟體Agisoft Metashaperup進行影像後製處理產製三維地形點雲。以1-2月為周期自2019年起至2021年為止，固定建構臺北市文山森林公園三維地形點雲，完整留下文山森林公園施工前、施工中、施工後之三維地形點雲。
接著利用多期三維地形點雲，其帶有空間座標以及顏色的特性，利用統計與分析找出施工區域點雲特性，包含鄰近點高度標準差、鄰近點前後期絕對高度差、鄰近點顏色，另透過實驗律定這三種特性鑑別是否為施工區域的閥值，最後透過閥值的鑑別找出三維地形點雲中正在施工的範圍，其中施工之定義依需求單位要求為伐樹整地、設施建設、加深開挖。
透過每期施工的範圍的偵測，當期施工面積得以量化，工程管考人員得以針對正在施工的區域聚焦考核，解決了以往二維平面影像不具有深度，無法正確辨識施工範圍的問題，省下了大量人員現地巡檢的成本，也加速了工程管考的審查。

This research intended to develop a detection strategy for construction region in hillside development, construction region includes site access for construction, grading works and felling trees. By taking aerial photo and placing ground control point to build 3D point cloud of a park construction. After building time series of 3D point cloud from 2019 to 2021, recording whole terrain state among construction period. Construction region can be detected and monitored after analysis twenty period of 3D point cloud.
Defining three types threshold values with properties of 3D point cloud, include geographic coordinates and color. Threshold types is defined as relative neighbor height, absolute neighbor height and color of neighbor. Distinguishing each point from raw to constructed region by threshold values. Once the status of point can be confirmed, construction region can be located and quantify size, hot spot of construction between period of point would be found out, construction region and area are valuable information for supervisor and achieving a cost reduction of patrol inspection.

1. Turner, D., A. Lucieer, and C. Watson, An automated technique for generating georectified mosaics from ultra-high resolution unmanned aerial vehicle (UAV) imagery, based on structure from motion (SfM) point clouds. Remote sensing, 2012. 4(5): p. 1392-1410.
2. Snavely, N., S.M. Seitz, and R. Szeliski, Modeling the World from Internet Photo Collections. International Journal of Computer Vision, 2008. 80(2): p. 189-210.
3. Lucieer, A., S.M.d. Jong, and D. Turner, Mapping landslide displacements using Structure from Motion (SfM) and image correlation of multi-temporal UAV photography. Progress in physical geography, 2014. 38(1): p. 97-116.
4. Campbell, N., G. Vogiatzis, and C. Hern. Using multiple hypotheses to improve depth-maps for multiview stereo. in The 10th European Conference on Computer Vision. 2008. Berlin: Springer.
5. Furukawa, Y. and J. Ponce, Accurate camera calibration from multi-view stereo and bundle adjustment. International Journal of Computer Vision, 2009. 84(3): p. 257-268.
6. Furukawa, Y. and J. Ponce, Accurate, dense, and robust multi-view stereopsis. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2010. 32: p. 1362–1376.
7. 農林航空測量所. 專業製圖介紹 http://www.afasi.gov.tw/cate. 2017.
8. 內政部國土測繪中心. 台灣通用電子地圖服務網 http://emap.nlsc.gov.tw/emap25/. 2017.
9. Liu, C.-C., et al., Rapidly responding to landslides and debris flow events using a low-cost unmanned aerial vehicle. Journal of Applied Remote Sensing, 2015. 9(1): p. 096016-096016.
10. Snavely, N., S. Seitz, and R. Szeliski, Modeling the world from internet photo collections. International Journal of Computer, 2008. 80: p. 189-210.
11. Sona, G., et al., Experimental analysis of different software packages for orientation and digital surface modelling from UAV images. Earth Science Informatics, 2014. 7: p. 97-107.
12. Lowe, D., Distinctive image features from scaleinvariant key points. International Journal of Computer, 2004. 60: p. 91-110.
13. Triggs, B., P. McLauchlan, and R. Hartley, Bundle adjustment: A modern synthesis. Vision Algorithms: Theory and Practice, ed. Z.A.a.S.R. Triggs B. 2000, Berlin: Springer.
14. Furukawa, Y. and J. Ponce, Accurate camera calibration from multi-view stereo and bundle adjustment. International Journal of Computer, 2009. 84: p. 257–268.