Solution Of Constraint Problem In 3D Visual Dynamic Measurement

With the development of the industry,the static three-dimensional measurement can no longer meet people’s production needs in certain industrial fields.Dynamic visual measurement effectively avoids the defects such as high measurement costs and tedious equipment maintenance of traditional non-destructive detection.Online detection,real-time control and other industrial scenarios.In the actual projects connected to the enterprise,the dynamic testing of important industrial parts such as seamless steel pipes and rail buckles,single-frame images need to measure at least 20 cm,and need to process more than 10 frames per second.Double/multi-eye requirement requires three-dimensional matching,grating scanning needs to be solved,the amount of computing is large,and the measurement speed cannot meet the demand.Therefore,this article set up a multi-line structure light single-eye visual measurement system based on the direct triangle method,and it is performed.Improvement provides certain solutions to the issues such as improper selection of light stripe patterns in the actual application scenario,difficulty in matching light stripes,and unable to recover the overall surface shape of the object.The main work content of this thesis is as follows:(1)Single multi-line structured light measurement model.A camera imaging model was constructed and the camera parameters were solved.The task and principles of light plane calibration were explained,and the light plane equation was solved.The calculation method of the actual measurement object was proposed,and the system was tested by the steel pipe and rail buckle as an example.The hardware test system was built with a projector,an industrial camera,and a laptop,and the software module was compiled and optimized by the Open CV library and QT framework to achieve the entire process of measurement system from the early stages to the later measurement.(2)Evaluate light stripes cover.When the surface shape of the object is complicated and the change is too large,some light stripes will be blocked and cannot be obtained by the camera.The error rate and loss rate of the data will be increased,which will affect the subsequent processing and analysis.The concept is to estimate the obstruction of the light stripes,so as to adjust the light stripe pattern,and develop the visual auxiliary software to simulate the cover..(3)The extraction and matching of the center point of the light stripe.Discuss the main optical stripe center extraction method in detail.Through experimental comparison,determine the extraction algorithm that is most suitable for multi-line structure light dynamic measurement,and improves.In this study,the central point matching of the light stripes does not refer to the matching of multiple points in multiple pictures,but refers to the pixels of the light stripe center point of the light stripes extracted in the single image.Three matching methods for different measurement needs and measurement scenarios are analyzed for different measurement requirements and measurement scenarios,and analyzed their respective scope of applications.(4)Restore the shape of the whole area.By extracting the utilization and data density of single images by extracting the center point of the dark stripes and the edge of the stripe,the B-Spline curve curved surface is introduced,and the node vector is used to use the cumulative chord length method.The scanned data point is generated continuous and smooth curved model,which improves the data density and restores the whole picture of the measured object.