3D Visual Contour Based On Projected Point Cloud Measurement Technology Research

With the rapid development of visual measurement technology,binocular vision 3D reconstruction technology has attracted widespread attention from scholars and engineers,and is currently widely used in spacecraft,unmanned driving,3D printing,robotics,intelligent manufacturing and other fields.Compared with traditional vision measurement,structured light-based vision measurement has certain advantages in terms of accuracy,robustness,and realtime performance.Therefore,this paper studies the structured light binocular vision three-dimensional contour measurement technology.First of all,the structured light dual-vision 3D measurement system in this article is composed of two Daheng industrial cameras and supporting lenses,a DLP4500 projector,supporting cables and a computer.In the three-dimensional measurement,the system in this paper uses the projector to actively project the four-step phase shift coding pattern and the Gaussian random speckle coding pattern,and sends the edge trigger signal through the trigger interface of the projector to trigger the binocular camera to collect the image synchronously,and the binocular camera shoots.Three-dimensional reconstruction of the object is performed after the image.In terms of structured light coding,in order to take into account the measurement speed and measurement accuracy,this paper combines the four-step phase shift method and Gaussian random speckle to design a hybrid coded structured light suitable for this system.Subsequently,this paper studies the ideal binocular vision measurement system model,analyzes the causes of camera distortion,and uses Zhang Zhengyou calibration method to calibrate the binocular camera to obtain the internal and external parameters of the binocular camera.Secondly,on the basis of four-step phase shift decoding and region-based matching algorithm,this paper proposes a phase-based NCC stereo matching algorithm for the measurement scheme of this paper.The algorithm implementation process is to first unwrap the phase of the left and right four-step phase shift projection patterns,use the left unwrapped phase image as the reference image,and the pixels in the image as the reference point,and search for the same phase on the same polar line from the right unwrapped phase image.As the point to be matched,the NCC correlation coefficient is calculated in the Gaussian random speckle image in the 5×5 area centered on the point to be matched and the reference point respectively to determine the best matching point,and finally traverse Repeat the above steps for all pixels of the reference image to complete the stereo matching.At the same time,in order to improve the calculation speed of the program,the two-dimensional search is reduced to onebit search by stereo correction,and then the search range is reduced from the length of 750 pixels to 200 pixels through the parallax range constraint,and the repetition value is reduced by saving the correlation coefficient.Based on the amount of calculation during region matching.Finally,in order to obtain high-quality point cloud data,the nearest neighbor algorithm is used to eliminate outliers.The prototype system built in this article is then used to measure the relative distance between different surfaces of a specific workpiece.The final experimental results show that the prototype system can achieve a plane geometric accuracy of 0.05 mm and a distance measurement accuracy of 0.13 mm at a working distance of 0.5m,which satisfies the system design.index.