Study On The Complex Profile’s Unambiguous3D Reconstruction And Its Related Algorithm In Reverse Engineering

Reverse engineering is a technology of convert the measured material object data tothe digital information, the key technologies including data acquisition, data processingand model reconstruction. With the development and mature of reverse engineering, inrecent years, many engineers and technicians have applied it to medicine field. Its mainapplications is reconstruction the three-dimensional model by the CT data, and get theSTL format file in order to make or produce the human skeleton or soft tissue model byusing the RP technology. So how to convert the complex profile information to threedimensional image information and display it is a critical problem. Two dimensionalimages visualization consist of a series of theories, methods and techniques, applyingcomputer graphic theory, image processing technique, computer vision theory andhuman-computer interacting technique to transform the two dimensional image data tographics and images, which can display on computer screen for interactive processing.In this paper, based on the two dimensional images such as CT data and its relativetheory is studied deeply, we main gives thorough study on the method for constructing3D model by2D images such as image preprocessing, consistency of topologicalrelationships, mesh simplification and other key technologies. The main work is asfollows:(1) The acquisition and preprocessing of two dimensional images. First, weanalyzed the basic principles of CT and its equipment constitution, settled the problemsof reading CT data and displaying it on computer and the image histograms isprogrammed and realized. Second, in order to solve the image noise problem in imagecapture and extraction, based on analyzing filter algorithms, we programmed andrealized Gauss noise filters and removes Gauss noises and impulse noises effect. Last,we studied segmentation algorithms, and programmed and realized the thresholdsegmentation algorithm and edge growing segmentation algorithm, laid the foundationsfor three-dimensional model reconstruction.(2) A new law is proposed to solve the topology ambiguity. Based on studying thetopology ambiguity theory and algorithms deeply, we proposed the decision law. Thenew algorithm introduces a cross section, and if two nodes of the same sign arenon-separated or joined inside a cube, they must be joined in this cross section. We fondout the tmand following the result the new decision law have been presented. Advantageof the algorithm is that: first, the law is available at any case; second, this method hasthe characteristics of less calculation, high velocity, high efficiency and etc.(3) A new mesh simplification is presented. The mesh models reconstructed by MC algorithm have huge grids. In this section, based on studying the existing meshsimplification algorithm deeply, we presented an improved method. The new methodbased on the classical algorithm QEM, using Gauss curvature, we define the concept ofcurvature factors of collapsing edge and embed it into the original Garland’s quadricerror metric. After adopting new methods, the cost of edge collapsing and the locationof new vertex would be both restrained by distance and surface geometric change, thedifference of high curvature and low curvature will be gradual increasing, this embodythe priority of low curvature, in this way, the vertices which have the sharp features willbe not removed. The experimental results show that the new method can reserve quiet anumber of important shapes features and reduce visual distortion effectively at lowlevels of detail.(4) A medical images reconstruction system based on the improved MC algorithmis developed. Current there are no three-dimensional model reconstruction commercialsystems on the domestic market. In order to conduct tests for the rationality and validityof above method and take the further research helpful, a medical images reconstructionprototype system based on the improved MC algorithm is developed, which includesmedical image pre-processing, segmenting and reconstructing modules. Examples of3Dreconstruction can keep the topology coherence, and represent the tissues or organs inrealistically as well. The reconstruction models can be exported and feed to a rapidprototyping machine to apply for producing organ and so on. If the system be give someimprove in future time, it can be popularized to use.