Research On The Technology Of Binocular Stereo Vision Measurement

After the appearance of Industry 4.0,the field of artificial intelligence has become a hot topic for researchers.Binocular stereo vision technology is also attracting more and more attention.The technology is widely used in the fields of industry,aerospace,and medicine.Computer vision technology is that using computers to simulate the functions of human vision and realize the perception of surrounding scenes.Binocular stereo vision is an important direction of computer vision research.Two identical cameras capture images in the same scene at different locations,and then the disparity map is obtained by stereo matching of these two images.Finally,the 3d depth of the actual scene is calculated based on the disparity map information.The technology of Binocular stereo vision measurement mainly includes image acquisition,camera calibration,image preprocessing,stereo matching and 3D information calculation.This article mainly focuses on camera calibration,stereo matching algorithm,and the calculation of 3D information.The main contents are as follows:1.In camera calibration,Zhang plane calibration method is used,and adopting the Matlab calibration toolbox to obtain the intrinsic and external parameters information of the two cameras,than making the stereo calibration.2.In stereo matching section,the theory of stereo matching algorithm is introduced firstly.A novel matching calculation method which color information combines with an improved Census transform is presented.In the cost structure,the combination of color information and improved Census transformation is adopted;In the structure of the support window,abandoning the color information similarity to build the window,and select the gray value similarity to construct the cross-based adaptive window.Finally,the images are tested online by using the Middlebury test platform,this algorithm has expressed better performance.3.In the depth information section,20 pairs of images are acquired by the platform which are built by two USB Microsoft cameras HD-3000.The intrinsic and external parameters information of the two cameras are obtained by using the Matlab calibration toolbox.Finally,the depth information of the object in the actual scene is calculated by using the correlation function in the OpenCV3.10.Compared with the traditional methods,The average error rates are 8.20% and 33.8% when using traditional images and new images respectively,the experimental data show that the stereo matching algorithm is presented in this paper has a lower mismatch rate.In the measuring part,the accuracy of the algorithm is more than 95%,which proves that the algorithm can be applied to the practical engineering.