基于机器视觉的巷道变形实时监测与预警系统设计

Design of Real-time Monitoring and Early Warning System for Tunnel Deformation Based on Machine Vision

采用双目视觉进行巷道变形实时监测,包括图像获取、图像处理、数据传输、图像分析、图像理解、数据分析。为提高图像处理精度,通过Shi-Tomasi算法与迭代算法完成亚像素级定位,并通过张正有标定法与非线性算法完成相机标定。基于极限约束对三维坐标进行重构,最后根据围岩变形量、变形速度、变形加速度确定预警准则,从而在保证正确预警的基础上降低虚警率。

Binocular vision is used for real-time monitoring of tunnel deformation,including image acquisition,image processing,data transmission,image analysis,image understanding and data analysis.In order to improve the accuracy of image processing,Shi-Tomasi algorithm and iterative algorithm were used to complete the sub-pixel positioning,and Zhangzhengyou calibration method and nonlinear algorithm were used to complete the camera calibration.The 3 D coordinates were reconstructed based on the limit constraint,and the warning criteria were determined according to the deformation,velocity and acceleration of surrounding rock,so as to reduce the false alarm rate on the basis of ensuring the correct warning.