Research On The Key Technologies Of Discrete Point Clouds Process

Points of cloud have errors of manufacturing, measurement, assembling and abrasion, the Three-dimensional Surface Model Technology of reverse engineering has to make researches on how to recover hidden geometric and topological information, and construct constraint. The major problem is how to combine with the existing physical object and the parameter to get out its initial design parameter, to elaborate the rationality of its existence and to boost the design level.This paper studies the problem of the discrete points of cloud process technology. That is, how to reconstruct mesh or surfel model to achieve its geometry property, and how to get the basic dimension and the constraint for CAD model of reverse engineering.The main researches are as follows:1. Perform pre-registration to the CAD model and the abrasive model with global registration based on the points of cloud data segmentation and the corresponding technology of the controlled point sets. Utilize the data segmentation, the distance of discrete points of cloud and the normal filter for the precise registration of the two models to recognize the damaged region. Based on the comprehensive analysis of effects on constraints of the multi-views registration error-chain such as performance, manufacturing cost and process ability, the error-chain of multi-views registration is constructed and a sensitivity sequence analysis algorithm is used to determine the measurement plan. The weighted average algorithm for multi-views registration is also presented.2. The computation method of non-manifold mesh model to be mesh solid model is presented. Moment of inertia and product of inertia about the elementary cell is deduced. Geometric properties of the three-dimensional model based on tetrahedron and parallel axis is studied. Through researches on complex solid model of the surfels model from the construction principle of the model to the geometric properties, this paper proposes an algorithm for surfels model from Boolean operation and CSG representation to its geometry property and discusses the accurate method of surface fitting for Boolean operations to resolve calculation of boundary intersection without any topological information.3. This paper discusses that utilizing the technology of data segmentation and feature line extraction reorganizes the scattered data points to layered structure, and keeps the features and reduces data redundant. The paper proposes to apply slicing to recognize features of the hole and construct partial surface matches. Adopt the method of getting points from the line to repair the hole. Based on the definition and classification of Variation geometric constrains, the generation and kinematics model of variation geometric constraints network, the existing physical object and the parameter of discrete points of cloud are combined. Making use of basic feature, associated feature and derived feature to construct the topological constraints and to seek the initial design parameter, establish the constraint equation of initial design parameter retrieval.In a word, this paper proposes new ideas and algorithms to each key procedure in discrete points of cloud information process, and presented new means and new techniques for reverse engineering in this dissertation. It indicates that they have wide applications in computer aided design and computer graphics.