自动判别基坑支撑位置的图像识别方法

Identifying Support Positions of Braced Excavations Automatically using Photogrammetry

摘要:

提出了一种基于设计信息的基坑支撑位置图像识别方法。首先，在室内相机标定的基础上对采集到的基坑图像进行镜头畸变校正；然后，利用控制点法确定相机在真实世界坐标系中的位置和姿态（以下简称位姿）；根据针孔摄像机模型成像原理将支撑区域投影到图像上；再对形成的支撑图像区域进行特征提取，使用颜色分量关系对图像特征进行分析，以达到判断支撑是否施工完毕的目的；对基坑的所有支撑进行循环判断，最终得到当前基坑施工的支撑位置。与传统数字照相识别方法相比，能有效避免复杂施工现场环境的干扰；最后，给出了一个基坑支撑位置识别的工程实例。结果表明，基于设计信息的识别基坑施工过程支撑位置的方法是有效可行的。

Abstract:

A design information-based photogrammetric technique of identifying support positions of braced excavations is presented. Firstly, the acquired image is rectified based on laboratory camera calibration parameters. Then, the position and orientation of the camera in the real world are calculated using the control points method. Support regions designated in the design are projected to the image according to the pinhole camera model. Features of the support regions are extracted from the image and analyzed using a proposed color component relationship to determine whether the support is constructed or not. Finally, all support positions can be determined according to design information and their positions can be identified one by one. This technique can effectively avoid interference caused by complex construction environment. A case study of this technique in a braced excavation project is also presented. The application shows this technique is practicable and effective.