Research On Displacement Measurement Method Of High-speed Railway Bridge Based On Computer Vision

In recent years,China's high-speed rail construction has accelerated into a network,and high-speed rail has become the primary choice for people to travel.The construction of high-speed railway generally adopts the way of "lead by bridge",so bridge structure takes up a large proportion in high-speed railway line.With the impact of high-speed railway on the bridge,the damage caused by bridge structure gradually accumulates,which will cause serious safety accidents.In order to ensure the safe operation of HSR,dynamic displacement of HSR Bridges should be monitored or checked regularly.The traditional contact measurement method is of low efficiency,high cost,construction difficulty,this article adopts the method of based on commercial camera capture image,based on the two kinds of design optimization based on feature point detection algorithm,realize the dynamic displacement of the bridge structure of non-contact measurement,the method has the characteristics of efficient,rapid,accurate,and the main research contents are as follows:(1)Study the practical measurement method of computer vision.For there are only three kinds of computer image acquisition mode were analyzed,on the basis of using monocular vision measuring bridge structure dynamic displacement measurement in the form of actual physical distance and pixel distance proportion relationship,determine the relevant conversion coefficient calculation formula,on the basis of the traditional visual target summary is suitable for the target design of civil structural displacement measurement principle,put forward a new type of circular targets and applied to the actual bridge model the dynamic displacement measurement.(2)Theoretical research on two methods of bridge displacement measurement identification.Purify and optimize multi-constraint feature points based on SURF feature point algorithm,effectively eliminate mismatched points,and use this method to locate and extract ROI.The optimized design was carried out for Harris corner detection algorithm.Forstner operator and singular value decomposition were adopted to eliminate the error points.The elliptic center was fitted with the least square method and the target center was precisely located.(3)Based on the laboratory measurement of dynamic displacement of the bridge model,the two algorithms optimized in this paper were used to measure the structural displacement respectively.The results were compared with those of the traditional displacement meter to verify the feasibility and accuracy of the two methods,and the corresponding evaluation and error analysis were carried out.The results show that the two algorithms can meet the requirements of measurement accuracy and provide theoretical support for the subsequent engineering practical application.In this paper,commercial cameras are used to collect dynamic displacement images of Bridges,and two algorithms are used to identify dynamic displacement of Bridges and conduct experimental verification.The results show that both methods have high accuracy and meet the needs of practical engineering measurement,which can provide reference for subsequent similar work and practical application.