Dynamic 3D Measurement Method Of Rotating Object Based On Line Structured Light

The structured light method is a 3D measurement technology based on optics.The existing research is mainly for the 3D measurement of static objects,which is of great significance to the protection of cultural relics,medical operations,and industrial component manufacturing.But for high-speed rail hubs,rotors of air preheater,and other dynamic rotating objects with high demands on measurement accuracy and large downtime losses,3D measurement technology has not made great progress.The research focus of this paper is to realize dynamic 3D measurement ofrotating objects.Based on line structured light method and stereo vision,this paper proposes a 3D measurement method that does not interfere with the running state of the object,is simple to operate,and does not require axis calibration.The measurement equipment is mainly composed of a line laser sensor and multiple binocular cameras.The main research contents of the measurement method in this paper are as follows:(1)Stereo vision 3D measurement: The high accurate result of stereo vision calibration is the basic guarantee of measurement accuracy.This article uses Zhang's calibration method to calibrate the internal and external parameters of the camera.Then the image stitching method based on homography provides the necessary parameters for the circular stitching,and this article optimizes the homography based on loop closure detection,which effectively reduces the cumulative error of long-term measurement.Image stitching without depth information will cause the position deviation of the spatial point.This paper fits the rotation plane where the axis is located,and uses it as a reference frame for spatial point correction.(2)Dynamic 3D measurement method of single measurement unit: First,use adaptive threshold method to extract the center of the line laser from the collected image,and use Gaussian filtering and nearest neighbor interpolation to smooth the fluctuations and discontinuities of the laser line with sub-pixel precision.Then calculate the spatial points based on stereo vision and laser features,and use the optimized homography to stitch these spatial points in a circular shape,thereby converting the measurement results to the correct spatial position,and obtaining the circular point cloud.Because the main optical axis of the camera cannot be strictly parallel to the rotation axis of the object,and the circular stitching lacks depth information,the 3D measurement results will be umbrella-shaped.In this paper,the rotation plane is used as the reference system to align and correct the umbrella-shaped points after circular stitching,which effectively eliminates the deviation of the spatial points.(3)Dynamic 3D measurement registration method of multiple measurement units: Due to the limitation of the camera's field of view,multiple measurement units are usually used for highprecision 3D measurement of large-scale objects.In this paper,the intersection area of the two adjacent measurement units is fixed,and the point cloud obtained in the intersection area is ringshaped.Based on this feature,improved ICP algorithm is proposed in this paper,and greatly reduces the amount of calculation and improves the registration accuracy.And finally convert all point cloud results to the world coordinate system to complete the dynamic 3D measurement of the object.The dynamic 3D measurement method has been compared with Kinect V2 and Ciclop(an open source solution for 3D measurement)in real-time and accuracy.And the physical measurement of the rotating object with complex texture is carried out.The experimental results show that the dynamic 3D measurement in this paper has a better performance in real-time,accuracy and restoration of the texture details of the object.In addition,the improved ICP algorithm is also verified,which proves that the dynamic 3D measurement method in this paper can perform high-precision 3D measurement on large-scale rotating objects.