High-Resolution Three Dimensional Meas-urement Method For Extendable Surfaces

Precision measurement represented by dimensional metrology is the foundation of precision machinery manufacturing.The emerging requirements in the manufacturing field drive the innovation of precision measurement principles and methods.As a cutting-edge field of dimensional metrology,the point cloud requires highly in measurement methods,accuracy,efficiency,and data density.Constructing the spatial structure relationship between the coordinates and dimensional features,three-dimensional(3D)surface measurement based on high-density point cloud has solved the problem of complex surface measurement on large-scale space,fulfilled the application requirements in industrial manufacturing,and become a development direction of dimensional metrology.Continuous high-density point cloud is the core of extendable surface highresolution 3D measurement,which breaks through the relative static constraints of the sensor and the measured objective and overcomes the limitation of the conventional 3D space measurement range.Requiring more highly the continuity of the measurement process and extensibility of measurement space,extendable surface high-resolution 3D measurement has a wider application prospect in rail transit and industrial assembly line manufacturing.Limited by the measurement principles of sensors,the current 3D surface measurement methods hardly obtain high accuracy point cloud in a continuous motion process and in extendable measurement space.In this dissertation,extendable surface high-resolution 3D measurement methods are lucubrated from high accuracy point cloud acquirement.The key technologies are researched,including the image matching strategy,the point cloud optimization,and calibration for line-scan cameras.The specific contents are as follows:1.A high accuracy point cloud acquirement method based on a line-scan continuous measurement model is proposed.The line-scan continuous measurement model is established.The line-scan visual measurement theory and continuous movement state are fused.The in-motion point cloud with high accuracy is measured by multiple linescan sensors whose acquisition time is asynchronous.The measurement error of the line-scan continuous measurement model is analyzed.2.The matching strategies for line-scan images are proposed.In spatial dimension,the one-dimensional background-normalized Fourier transform is utilized to solve the problem of insufficient features in unidimensional images and ambient light interference.In time dimension,the epipolar constraints of line-scan cameras are constructed to complete the accurate matching of extendable line-scan images.3.The local point cloud accuracy optimization method based on bundle adjustment for an in-motion target is designed.The bundle adjustment for in-motion target based on the Gauss-Newton iteration is used to optimize the instant motion state of point cloud.The accuracy of contour stitching and one-shot measurement is improved.The optimization of in-motion the point cloud is implemented.4.A calibration method for line-scan cameras using a stereo target with hollow stripes is proposed.By the theory of cross-ration invariance and harmonic conjugate,the eccentricity errors are corrected and the camera’s calibration process is simplified.The high-accuracy calibration for the measurement systems based on line-scan cameras is realized.In this dissertation,the measurement methods of the continuous high-density point cloud are researched in-depth.Measurement systems of dual line-scan cameras and triple line-scan cameras were built.The feasibility of the measurement theory was verified.The accuracy of measurement accuracy was improved.The advantages of the measurement system were highlighted.Finally,the extendable high-resolution 3D surface measurement was implemented.