Application Research On Analysis, Segmentation And 3D Reconstruction Of 3D Point Cloud

At present,the hardware and software of the computer have been developed by leaps and bounds,and the three-dimensional reconstruction technology has made considerable progress compared with the past.The purpose of 3D reconstruction is to create a 3D model for computer processing and analysis for objects or scenes.Currently,the raw data obtained by the scanner device or the image-based three-dimensional reconstruction method is three-dimensional point cloud data.However,in the related applications of 3D printing,virtual reality,digital city,intelligent transportation,etc.,the polyhedral surface model or CAD solid model of the target object is required.Constructing a 3D model from a 3D point cloud is one of the key technologies for 3D reconstruction and the core issue of reverse engineering.In general,narrow threedimensional reconstruction refers to the process of constructing a three-dimensional solid model from a three-dimensional point cloud.In practical applications,since a three-dimensional point cloud collected by a method such as three-dimensional scanning and image processing is complex data containing a plurality of objects,Therefore,segmentation of point clouds in units of objects is a necessary preprocessing method in the process of three-dimensional reconstruction.At present,point cloud segmentation processing mainly relies on relevant professional software to complete manually,and there are problems such as high processing cost and low efficiency.This thesis mainly studies the automatic segmentation method of complex point cloud data and its application in 3D reconstruction.This paper proposes a segmentation method based on planar model,which uses the idea of stepwise segmentation.Firstly,fit the plane model in the 3D point cloud,and count the position of the 3D point cloud in the plane model.The plane model with a three-dimensional point cloud on one side of the plane is reserved as the boundary plane model of the point cloud,and the boundary plane model is separated from the initial point cloud,thereby generating a three-dimensional point cloud with obvious gaps.Then use these gaps to classify 3D point clouds.Finally,the plane model is fitted in each type of point cloud and paralleled with the boundary plane model.As the cutting plane of the point cloud,the segmentation method under the object stacking condition is realized.This paper proves that the method has certain effectiveness through experiments.