近景攝影測量於施工進度監測之應用

Application of Close-range Photogrammetry for Construction Progress Monitoring

曾昱晨(Yu-Chen Tseng)

指導教授 ： 徐百輝

國立臺灣大學/工學院/土木工程學研究所/碩士(2013年)

https://doi.org/10.6342/NTU.2013.01813

摘要

營造業使用三維3D視覺化的技術輔助施工進度監測已逐漸成熟，但實務上多用於與業主溝通等單純視覺化展示的工具。規畫設計時的模型未能配合實際現況進行更新，導致無法達到進度監測的目的。本研究使用近景攝影測量的技術，定期拍攝施工現況的影像，並將預先規畫設計的三維模型投影到二維影像中進行套合，藉由不同時期的模型投影來辨識建築物件的進度，輔助專案人員瞭解工程實際進度與預定進度的差異。
本研究欲發展一套半自動視覺化進度監測流程並應用在實際案例中。視覺化進度監測流程如下：(1)影像方位重建：將所拍攝的施工現場影像建立其方位，每一張影像對應到的相機位置。(2)模型投影：將3D模型藉由共線式以及影像的方位參數投影進影像之中。(3)模型重建：將投影的模型利用最小二乘法套合至影像中的建物上，並輸出模型參數。(4) 影像進度辨識方法：將不同時期的模型投影至拍攝的影像之中，依據本研究建立的辨識邏輯來判斷進度是否符合。最後利用此進度監測模式，進行實際案例的驗證。
經過實際案例的驗證，透過視覺化進度監測，即可半自動比對施工現況與預定進度的差異，並提供及時的進度資訊，幫助專案人員或者是第三方的監測單位，更快瞭解實際狀況，並期望能夠一步一步達到自動化進度監測的目的。

關鍵字

近景攝影測量 ； 工程影像辨識 ； 視覺化進度監測 ； 營建自動化

並列摘要

In construction industry, 3D Visualization technology has come into being in monitoring construction progress. However, recent 3D models are mainly used as a visual displaying tool for contractors communicating with others. Lacking of periodically real-time updates constrains the pre-planned models to serve as monitoring purposes when the techniques put into practice. To overcome the limitations of existing applications, this research focuses on the monitoring technique using as-built photographs matching with images taken from the 3D as-planned model in different construction status to recognize the actual working schedule. This approach enables the project managers to have a realistic understanding of the deviation between the pre-planned schedule and real-time construction progress.
The purpose of this research is to develop a visual progress monitoring system that can be applied to practical projects. The chronological use of the visual progress monitoring system includes: (1) Reconstruction of the Image's Orientation: Obtain the orientation of an image from a worksite. Each image will be able to accurately depict the orientation of the camera. (2) Projection of a model: Using the Collinearity Equation and the orientation of the image, project the 3D model onto the worksite's image. (3) Reconstruction of the Model: Match of projected model to it's corresponding component at an worksite using Least-Squares Estimation, and print out the model's parameters. (4) Method of Visual Progress Monitoring: Project individual models taken from various times as the work progressed onto the worksite image. Using the logic of recognition developed by this research to conclude whether the progress at the worksite is up to par with the predetermined schedule. Finally apply this technique at an actual worksite.
It has been experimentally determined that using the visual progress monitoring system, the worksite's progress can semi-automatically compare the actual progress to the planned progress. Thus, the system can provide information regarding the worksite progress in a timely manner, and thus assist professional or a third-party monitor in understanding the actual progress. This research aspires to elicit further research and eventually develop an automatic progress monitoring system.

並列關鍵字

close-range photogrammetry ； construction image recognition ； visual progress monitoring ； construction automation

參考文獻

蕭琮輝，2009，應用影像辨識技術於營造業工程進度管控之研究，碩士論文，國 立台灣大學。

Chio, S. H., 2001. A Practical Strategy for Roof Patch Extraction from Urban Stereo Aerial Images Doctoral dissertation, Ph. D. Dissertation, National Cheng Kung University, Taiwan, ROC.

林文棋，2001，半自動化建物萃取之建物模型建置與操作，碩士論文，國立成功 大學

王聖鐸，2003，以浮測模型理論萃取三維空間資訊-以建物重建為例，博士論文， 國立成功大學

張智棋，2004，4D施工計畫之資訊管理研究，碩士論文，國立成功大學。