Research On Key Technology Of Roadbed Subsidence Displacement-relay Camera Network Videometrics

Tiny roadbed deformations of infrastructure, such as, railways, bridges, tunnels, building, may cause cracks, which may affect human daily life. If the deformations go on, it will lead to slides or collapses and brings about serious property damage, loss of life and personal injury. So it is necessary to research and develop a monitoring method and system, which can automatically measure the deformation of roadbed at real-time and prewarn the potential safety hazard. Overcoming the weakness of existing measuring methods, “roadbed subsidence displacement-relay camera network videometrics” is capable of measuring the subsidence, pitch and rotation of railroad beds, basing on the “fixation constraint” of “multi-headed camera” and the “displacement-relay”. In this paper, we research on the key technologies of “roadbed subsidence displacement-relay camera network videometrics”.For the purpose of improving the automaticity of videometrics, firstly, we research on full automatic calibration. Automatic calibration is implemented by automatic extraction, automatic matching and high precision calibration. We mainly focus our attention on automatic matching and bring forward the focal length coherence automatic matching algorithms and minimal residual automatic matching algorithm. N umerical simulation is implemented to proof the validation of proposed algorithms.This paper applies videometrics to monitor roadbed subsidence; designs measuring cell, displacement-relay chains and the monitoring system; formalizes a double-headed camera roadbed one-side subsidence monitoring model, on the condition of hypothesis and constraints; the validation of proposed monitoring method are proved by numerical simulation. Moreover, we analyze the influence of image error, level of displacement-relay and the position of reference points with respect to roadbed measuring accuracy. Basing on the design errors of multi-headed camera and initial errors of camera coordinates, we formalize the modified “double-headed” camera monitoring model and the generalized “multi-headed” camera monitoring model by using calibration results. Numerical simulation shows that the modified monitoring model can get better result.Simulation result of chapter 4 shows that the measurement error increases as the increasing of level of displacement-relay. By analyzing the matrix of coefficients, we find that the condition number of this matrix is too big, which causes big measurement error. Thus, we use principal component analysis algorithm to solve the monitoring model; Numerical simulation is implemented to verify the validation of this algorithms; comparing to least square algorithm, principal component analysis algorithm can solve this problem better.Physical experiment of roadbed subsidence measurement is designed and the corresponding monitoring system is implemented. Experimental result proofs the validation of “roadbed subsidence displacement-relay camera network videometrics”.What’s more, we can design different cooperative markers, multi-headed cameras and displacement-relay chains with respect to corresponding measuring circumstance and requirements. Therefore, this method has crucial theory value and broad prospect of being adopted to measure roadbed subsidence.