Research On Photographic Geometry Measurement System Based On Machine Vision

With the rapid development of China’s manufacturing industry,the composite functions required for the workpiece are very prominent,resulting in more complex workpiece contour shape and higher processing accuracy requirements.In the workpiece quality inspection,the traditional workpiece contour measurement method is inefficient and costly,while the conventional image-based measurement method has problems such as low measurement accuracy and poor stability,so how to measure the workpiece contour quickly and with high accuracy becomes a bottleneck problem for high-quality processing.In this thesis,we use machine vision-based photographic geometry measurement technology to detect the distance and angle between the double contour lines of a workpiece,including corner and edge extraction,sub-pixel positioning algorithm and double telecentric imaging,and apply them to the measurement of a typical workpiece.Firstly,a machine vision-based interactive measurement system for photogrammetric geometry is designed based on the contour of the workpiece and the required speed and accuracy index requirements for measurement.The hardware design of the system includes the selection of equipment such as camera,telecentric lens and light source,as well as the indirect line installation and position calibration of the hardware;the software design of the system contains functions such as template selection,batch measurement and data analysis.Secondly,based on the requirements of measurement accuracy,the corresponding series of algorithms are selected through the analysis and comparison of algorithms.In the edge and corner detection,three corner extraction algorithms and four pixel-level edge detection operators are analyzed and compared,and the Harris extraction algorithm and Canny detection operator with stronger interference resistance are selected;for the shortage of pixel-level detection algorithms in detection accuracy,an improved interpolation sub-pixel positioning algorithm based on Canny operator is proposed to precisely locate the coordinates of contour points to four decimal places The RANSAC algorithm with iterative effect is proposed to precisely fit the linear contour for the shortcomings of the least-squares fitting algorithm in eliminating interference noise,and the experimental data are compared and analyzed.Finally,the system is experimentally calibrated and the measurement parameters are synthesized and analyzed.Based on the analysis and modeling of the dual telecentric imaging method,the values of the distance between the dual profile lines of the workpiece and the pinch angle are calculated using the multi-sampling interactive measurement method.The experiments and error analysis of the measurement of the distance and angle of the linear profile of the workpiece at different positions in the axial direction show that the error range and repeatability of the measurement data meet the requirements of the index,which proves that the system has the characteristics of high accuracy,high stability and high speed.