Study Of The Recognition For Position Of Parts Based On Binocular Stereo Vision

With the rapid development of manufacturing technology,the industry gradually towards intelligence,automation development,the traditional industrial detection system flexible bottom,labor costs,in many ways already cannot satisfy the requirements of manufacturing.The addition of machine vision technology to the industrial inspection system can not only improve the automation degree of the system,but also reduce the cost and improve the working efficiency.Binocular stereo vision measurement technology is an important branch of machine vision.It has the advantages of high accuracy,fast speed and non-contact measurement.It is a research hotspot in measurement technology.In this paper,based on binocular stereo vision system of components of the three-dimensional measurement of in-depth research and experimentation,and calibration method for measuring algorithm was improved,3d measurement accuracy and efficiency,the concrete research content has the following several aspects:Firstly,the development status of binocular stereo vision in measurement and the research results obtained at home and abroad are introduced.According to the measurement requirements,the system software module and the hardware module are designed,the hardware and software are connected,and a set of binocular stereo vision system for the measurement of part position is formed.The measurement principle of binocular stereo vision system is analyzed and the mathematical model of binocular stereo vision system is established.Based on the mathematical model,the influence of system parameters on the measurement precision is determined.When the system is measured in three dimensions,the system must be calibrated,and the accuracy of the calibration results will have great influence on the measurement results.In order to accurately determine the calibration parameters of the system,the calibration method is studied firstly.In this paper,the calibration method of binocular stereo vision system based on HALCON is proposed,and thecalibration parameters of the system are obtained by using the homemade calibration plate to calibrate the system.Feature point extraction and image matching.Study of feature point detection algorithm,based on the characteristics of Harris algorithm and SIFT algorithm,to improve the feature point extraction method,puts forward the normalized image matching method based on SIFT algorithm,and experimental verification.According to the obtained feature point coordinates,the matching point parallax is obtained to complete the reconstruction of the three-dimensional point of space.Through the experiment,the two-dimensional graph is shown in three dimensions,and the three-dimensional coordinates of the parts are obtained.