Research On The Measurement Of Mechanical Size Based On Binocular Stereo Vision

Measurement of the size of the workpiece is an important content in the field of industrial production.The traditional measurement of workpiece size is mostly obtained by artificial measurement,which has the disadvantages of large artificial influence,low efficiency and random error.In recent years,with the development and application of new technology,workpiece size measurement based on binocular vision has been gradually applied in industrial production,and has gained increasing popularity due to its advantages,such as no need to contact,high accuracy and controllable equipment cost.Based on the hardware platform of the binocular vision system and the software platform of OpenCV and MATLAB,the measurement of the mechanical size of the workpiece is realized.The main contents are as follows:(1)Camera calibration.After introducing the related coordinate system and the imaging model of camera,the classical Zhang’s plane calibration method is used to complete the camera calibration work on the MATLAB software platform,and the camera’s internal and external parameters are obtained.(2)Feature point extraction.First,the famous SIFT operator in the field of feature point extraction is introduced.The extracted feature points are the extreme points in scale space,but uncertainly the "corners" that contain size information of the workpiece,the visual sense of the human eyes.Therefore,this paper proposes an improved SIFT operator which combines the advantages of SIFT operator and corner detection FAST operator.The experimental results show that the improved SIFT operator is able to extract the required feature points and measure the size of the workpiece.(3)Stereo matching and three-dimensional reconstruction of feature points.First,the feature points extracted from the left and right images are matched by the CN/SCN method.Then,the mismatched points are removed by using the pole line constraint rule.Finally,using the binary search method to select the desired matching points among the remaining points.After obtaining the required matching points,according to the binocular vision principle and the internal and external parameters obtained by camera calibration,the three-dimensional coordinates of the corners of the workpiece are obtained,and then the relative dimensions of the workpiece can be calculated.(4)Improved binocular vision measurement method.The classical size measurementmethod based on binocular vision has the defect of extracting too many feature points,which leads to the increase of time consuming for describing and matching the feature points,and makes the manual selection of the necessary feature points complicate and thus consumes too much time.Therefore,an improved dimension measurement method based on binocular vision is proposed,which extracts the workpiece from the background and then extracts and matches the features.Experiments show that the method can effectively reduce the number of feature points and the time of matching,improve the efficiency of screening and speed up the matching speed.According to the comparison between the size measurement results and the actual size of the workpiece,it is verified that the measurement method based on binocular vision is feasible and able to meet the required measurement accuracy of the general industrial site.