Point Cloud Splicing And Topology

If a leak occurs in the petrochemical equipment, it will cause very serious problems.Hence it’s quite necessary to check the safety of petrochemical equipment regularly byusing robot vision system. In order to make sure that robots can quickly find certaincomponent, there is a need to build the data model of special parts of petrochemicalpipelines. Point Cloud Model is a way to attach this goal. It can easily buildthree-dimensionalmodel through linking information (location, shape, etc.) of samplingpoints. This model data has a simple structure, which is easy to Storage and editing, makingit have the great advantage in the application.In this assignment, CCD camera is used to gather point cloud data of petrochemicalcomponents. There is a depth study on the pretreatment technology, point cloud splicingtechnology and point cloud topology relations of gathered data.Besides that, athree-dimensionaldata model of components is also built in this study. Followed are themain jobs that are going to be finished in this assignment:1.Conducting extraction operation of components on the depth-image that is shown onthe computer to get complete point cloud information. After that there is a need to dode-noising on the information to create8–neighborhood structure. Then additionalimplication on point cloud information of components can be done through curvatureestimation.2.Based on improved ICP algorithm, coming up with the method that the combinationof common points and ICP can splice scattered point cloud. According to the theory ofSkeletal Methodology, building point cloud topology relationship. Using the combinationof key points and boundary points to create complete special points. Matching them withthe boundaries to get boundary curves of shape features of component model.3.Based on point cloud cluster-cluster geometry and topology relationship and thetarget body semantic description method, establishing a complete three-dimensional pointcloud model library and conducting extraction of mechanical parts.