Research On Structural Spatial-Temporal Displacement Monitoring Method Based On Sequence Image

Sequence image can track the moving object by recording its spatiotemporal information with the advantage of efficiency,economy and noncontact,and has been widely used in public safety monitoring and protection,dynamic displacement deformation measurement.It is highly compatible with the high-precision online spatial-temporal displacement field measurement and damage identification demands of large structures,which is essential to solve the three major problems in existing large bridge health monitoring system:limited number of monitoring points,low accuracy,and computational efficiency due to high redundancy in feature extraction and matching algorithms.This dissertation is financially supported by national science foundation project Research on the Method of Obtaining the Bridge Deformation and Deducing the Structural State Based on Differential Analysis of Image Contour（51778094）,and project Evolution and Failure Process of Multiscale Analysis Method of Bridge Structure Based on Surface Shape of Spatial-temporal Data（51708068）,and carry out systematic research on basic theory,algorithm and test technology of lightweight monitoring of temporal and spatial dynamic displacement of bridge structures based on digital image sequence data.The main research contents of the thesis are as follows.The photoelectric energy imaging theory,image process and precision of motional objects are analyzed from the viewpoint of discretization,quantization,characterization from four dimensions of spatial,temporal and gray scale,and the thesis reveals the relationship of spatial,temporal and gray resolution and precision of machine vision measuring system.One-dimension subpixel image is constructed based on photoelectric energy imaging theory of each pixel,and the author develops constant gradient subpixel measuring model,from which the concept and computation formula of Theoretical Minimal Identifiable Motional Distance Δx_Tmin（TMID）,as well as Practical Minimal Identifiable Motional Distance Δχ_pmin（PMIMD）,are proposed.The component selection of industrial machine vision measurement system is systematically studied,including camera,data transmission interface,trace mark coding automation,auxiliary lighting and supporting optical devices.Selection,process,steps and implementation strategies for structural deformation measurement system are proposed.Subpixel algorithms of local structural motion tracking based on feature point extraction and matching and image correlation method are studied.Indoor Vibration test and synthetic image numerical simulation of indoor cantilever beam are carried out using template matching method,and local feature tracking accuracy of structure is verified.Based on prior photoelectric energy imaging theory,a two-dimension subpixel model is developed and reveals the energy mechanism of sub-pixel measurement.,and a subpixel structural deformation measurement scenario and model derived from energy area effect algorithm is proposed.Experiments show the precision advantage by comparing with wo dimensional brightness distribution sampling fitting subpixel algorithm,and the method is applied and verified in indoor thin-wall vertical beam vibration test.To solve the difficulties of calibration and error analysis in structure tempo-spatial deformation machine vision measurement,a forward measurement method and process are proposed.Ideal Edge Synthesized Image（IESI）reveals that pixel edge location error follows spindle shape distribution,and results show that the slope and intercept of edge equation determines the shape length,density and location of the spindle.Based on photoelectric sensor camera pixels record energy principle,three crossover and cut model are put forward for a single pixel and ideal edge,and pixel gray value of photoelectric sensor energy algorithm are put forward to generate Synthesized Subpixel Image of Ideal Edge（SSIIE）.Comparison between the ideal edge synthesis subpixel image and edge image from camera shows good accordance,which verifies the synthesized image model and algorithm.For complex scenarios structure deformation measurement by natural texture feature point extraction and matching algorithm,invalid feature points elimination,feature points density control,mismatching are three main problems,and priori structure knowledge are embodied as structural priori information（SPI）,geometric priori information（GPI）,location and deformation are put forward and introduced in structure image space segmentation method（SSSM）.Lightweight model for feature extract（LMFE）is obtained with prior knowledge image space segmentation method,and total feature search space is reduced about 83%.With component-based segmentation of total image,the overall feature extraction time was reduced by about 62.91%,with a large number of invalid environmental feature points and possible mismatched features being effectively eliminated.According to the geometric and deformation characteristics of large slenderness ratio structures,including beams,columns,cables,rods and piers,the concept,method and process of Slice Template Matching Method（STMM）are proposed as required by structures spatial-temporal continuous deformation measurement.The STMM method effectively solved the problem of limited number of measuring-point arrangement,distinction indiscriminability of mask match method along the beam length.Indoor experiment is carried out with a simply-supported box beam vibration measurement and a high precision lightweight spatial-temporal deflection field is obtained and verified.The concept and method of noise analysis and displacement anomaly data elimination are proposed based on the consistency constraints of multi-motion parameters.Numerical simulation and machine vision measurement of indoor beam with and without damage were carried out,and the full space-time curvature mode of the beam structure was extracted,with which beam structure damage was identified and verified.