Research On The Key Technologies Of Line Laser Point Cloud Data Processing

With the development of three-dimensional(3D)measurement and the continuous progress of computer data processing technology,line laser scanning technology has been widely used in industrial measurement,reverse engineering,and other fields.The line laser 3D scanning technology based on laser triangulation has the characteristics of high accuracy and flexible measurement.However,the initial point cloud data obtained by line laser scanning technology contain noise date and incomprehensive point cloud data when affected by object surface reflection characteristics,color characteristics,and scanning system errors,which lead to point cloud measurement errors.Therefore,preprocessing,registration and measurement is of great importance in point cloud data processing.The research contents of this paper are as follows:(1)An improved point cloud preprocessing based on the RANSAC algorithm is proposed considering the shortages of the Random Sample Consensus(RANSAC)algorithm,which are the requirement of setting the threshold value manually and the frequency of detecting wrong models when there are little internal points.The RANSAC algorithm is improved by using the median absolute deviation to automatically select the threshold value and set the stop iteration condition to fit the plane.Then,the point cloud containing the object is segmented according to the range of plane parameters,and the discrete noise points are removed using statistical filtering to obtain the target point cloud.The experiment proves that the improved method has a better pre-processing effect and,a wider application range,which provides good initial values for subsequent measurements.(2)A point cloud registration method based on image features is proposed for the problems of long registration time and low registration accuracy when the traditional 3D point clouds are partially overlapped.Firstly,the 3D points are mapped into the two-dimensional(2D)image by the line scanning topology of the point cloud.The Scale Invariant Feature Transform(SIFT)algorithm is used to extract image feature points,and three corresponding points are selected by the distance and angle.The rotation translation matrix between two point clouds is calculated according to the correspondence between the pixel points and the point cloud to realize the coarse registration.The point clouds after matrix transformation are simplified by voxel sampling,and the overlapping region is calculated.The point clouds in the overlapping region are used to realize iterative closest point precision(ICP)registration.Compared with the commonly used registration algorithm,the operation speed of the actual 3D point cloud data has tripled and applicable to the partially overlapping point clouds,which verifies the effectiveness of the method.(3)A method of workpiece geometry measurement based on point cloud is proposed aiming at the problem of low efficiency and poor precision of manual measurement of the workpiece.Taking the round hole workpiece class as an example,based on point cloud coordinates and image feature measurement respectively,based on the coordinate measurement is to transform the 3D point cloud to be parallel to the XOY plane through coordinate transformation,the height is obtained by the difference between point cloud and datum plane on the object surface,and the volume and surface area are obtained by triangulation.Based on the image feature measurement,the point cloud projection is mapped to two dimensions to form a binary image.The boundary length is obtained through image preprocessing,and the radius of the circle is obtained by using the constrained least square method.By measuring the standard gauge block,the measurement accuracy of the system is verified to be ±0.01 mm.The feasibility of the method is verified by using a circular hole workpiece.