Research On Measurement Technology Of High-Precision Two-group Binocular Vision System

With the increasing development of computer vision technology,non-contact measurement technology based on computer vision has been widely used in aerospace,automatic driving,intelligent robots,automated industrial production and other fields by virtue of its fast processing speed,high detection accuracy,ability to adapt to harsh working environments,and high degree of automation.Binocular or multi-mesh(greater than or equal to 3)visual measurement systems have higher measurement accuracy and stronger stability in large-sized structural rigid body measurement tasks than monocular visual measurement systems,and have more design results and more wide applications in practical applications.In this paper,a set of stereo vision measurement systems useing multiple groups of binocular vision controllers is designed,to ensure the technical problems of improving the measurement accuracy and data stability of the visual measurement system,ensuring the reliability of the visual measurement system used in the outdoor harsh environment,and ensuring the replacement of the binocular vision controller without calibration during the use of the replacement visual measurement system.The structural design ensures the repeat installation accuracy of each set of binocular vision controllers in the vision measurement system,and the installation accuracy is used to ensure the measurement accuracy when replacing the standby binocular vision controller to avoid repeated calibration process.The structure of the binocular vision controller housing is designed to ensure that its size,weight and sealing meet the technical requirements.The robustness of the algorithm of the vision measurement system is improved to ensure the reliability of measurement in different lighting environments.The characteristic patterns and feature point forms of cooperative targets are selected,and different target feature pattern forms are selected for different target identification algorithms and feature point position solving algorithms.According to the common feature forms,the original target target identification algorithm based on template matching and the feature point positioning algorithm based on corner point detection are compared with the target identification algorithm based on feature information screening and the feature point solving algorithm based on circular outline center of the circle of the existing design scheme,and the comparison results are experimentally verified to demonstrate the superiority of the new algorithm in the use scenario of the project background.In this paper,the environmental adaptability of the visual measurement system is studied,the environmental adaptability design is carried out for the "six requirements" that the measuring system needs to meet,the environmental factors that have a greater impact on the visual measurement,such as the measurement in the humid and hot environment and the measurement in the low temperature environment,the problem of frosting and the reduction of accuracy of the visual measurement system in the above two environments is designed,and the design scheme is experimentally verified,and the measurement accuracy of the visual measurement system is tested in different environments.