Research On Automatic Extraction Of Key Dimensions Of 3D Point Clouds In Mobile Phone Cases

With the popularization of 3D scanning equipment,it is easier and easier for people to obtain 3D data of objects.Three-dimensional data has broad development prospects in many fields such as surveying and mapping,unmanned driving,e-commerce shopping,cultural relics protection,reverse engineering,and industrial automation.Mobile phone manufacturers need to test the key dimensions of various types of mobile phone shells during product quality testing to ensure the yield rate.Commonly used detection methods for mobile phone shells are contact-type coordinate measuring machines and two-dimensional image measuring instruments.There are many tiny structures of the mobile phone shell,and there are many items to be tested.The coordinate measuring machine takes a lot of time and is inefficient;and the two-dimensional image measurement equipment can only meet the detection of the two-dimensional size,and the longitudinal size of the mobile phone shell is difficult For measurement,it is necessary to resort to line laser scanning equipment,but the test equipment has a complicated structure and high cost.In view of the above problems,this paper has carried out research by constructing a two-node three-dimensional imaging system,taking the three-dimensional point cloud of the mobile phone shell as the research object,and extracting from the two key issues of the feature boundary of the point cloud and the automatic measurement of the geometric dimensions of the mobile phone shell.The main work of this article is summarized as follows:The key technologies for preprocessing of scattered point clouds are studied,including: establishment of spatial topological relationship of point clouds,estimation of point cloud normal vectors,and point cloud denoising.First,the spatial topological relationship of scattered point clouds is established using Kd-trees,then the normal direction of sampling points is estimated using local K-neighbor principal component analysis,and finally the point cloud denoising is performed using a combination of statistical filtering,radius filtering and bilateral filtering.The point cloud characteristic point is comprehensively judged by using the average distance of the local point cloud,the angle between the normal vector between the point and the neighborhood point,and the point curvature.The method of judging the boundary point of the point cloud is analyzed,and a method of extracting the boundary feature points using three thresholds is proposed.Combining the distribution of the local point cloud,the normal angle between the point and the neighboring point,and the curvature information of the point are used as the basis for judging the boundary point.Quickly extract the characteristic boundary of the target point cloud area.According to the step size of the mobile phone shell,the radius and length of the camera hole,the plane size of the spacing of the positioning holes is extracted.Random sampling consistency is used to fit the reference plane and the step plane to calculate the spacing between the planes;for the plane class size,a method of mapping three-dimensional data to two-dimensional detection is used,and a mobile phone camera hole based on Hough transform is proposed.Dimension detection method;compare the measured point cloud of the simulated part with its standard point cloud,and the error distribution of the part,that is,the Hausdorff distance,is visually displayed using the color spectrogram.Based on the C ++ language and Qt tools,the software for measuring the size of the point cloud of the mobile phone shell was written.The software part includes functions such as fast loading of point cloud data,point cloud filtering,point cloud boundary extraction,point cloud plane fitting,and one-key measurement of column height.It realizes automatic plane fitting of interactive regions of interest,feature boundary extraction and one-key automatic measurement of step size.