Research On The Workpiece Positioning And Size Measurement Based On Binocular Vision

The positioning of workpieces and mechanical dimension measurement is an important research content in the field of modern industrial production.With the continuous development of robot technology,the method of manual measurement of workpiece size has been gradually eliminated due to its low efficiency,large randomness of error and vulnerability to human factors.It is replaced by a new type of measurement method.,that is,computer vision measurement.It has become more and more important because of its advantages of non-contact,good detection efficiency,high automation level and low cost,and it has become a very popular workpiece measurement method.Binocular vision measurement is a key research direction of computer vision detection,and its development has been paid more and more attention by researchers at home and abroad.The binocular vision measurement system is mainly composed of camera calibration,feature point extraction and detection,image stereo matching and three-dimensional reconstruction.In this paper,the positioning and dimension measurement of the workpiece is based on the binocular vision inspection system.Based on the in-depth study of each link of the binocular vision detection system,a new improved method was found to improve the measurement accuracy of the workpiece.In the aspect of camera calibration,based on the research of several popular calibration methods,according to the actual workpiece measurement requirements,the classic Zhang Zhengyou plane calibration method is selected,and the camera calibration operation is realized on the MATLAB software platform.Experiments show that the calibration method has high calibration accuracy for the camera and can meet the operational needs of subsequent processes.Aiming at the problem of feature point extraction and detection,this paper studies several common feature point detection algorithms,and analyzes the advantages and disadvantages of these algorithms through comparative experiments.Then,based on the above,an improved feature point detection method is proposed to ensure that the detected feature points have scale invariance and improve detection efficiency.Stereo matching of images is one of the focuses of the entire binocular vision detection system.Its matching accuracy is related to the accuracy of the final workpiece positioning and dimension measurement.In this paper,through the main processes of stereo matching,such as the selection of similar metrics,the selection of stereo matching primitives and the determination of spatial search strategy,the required matching pairs are accurately detected.In addition,this paper has improved the traditional binocular vision measurementmethod to a certain extent.Firstly,the problems of the structural error of binocular vision system and the influence of system calibration accuracy on the accuracy of object measurement are analyzed.The related factors are optimized and an ideal detection environment is found.Then,based on this,the workpiece is quickly positioned and measured by separating the workpiece from its background image and then separately processing the separated workpiece image.Finally,it is confirmed by experiments that this method has a certain degree of improvement in the measurement accuracy and efficiency of the workpiece compared with the traditional binocular vision measurement method.