斡泻朴嚎.PDFVIP

Vol.17 No.3 J. Comput. Sci. Technol. May 2002 Two Accelerating Techniques for 3D Reconstruction LIU Shixia (   ), HU Shimin (
  ) and SUN Jiaguang (  ) Department of Computer Science and Technology, Tsinghua University, Beijing 100084, P.R. China E-mail: liusx@ncc.cs.tsinghua.edu.cn; fshimin,sunjg g@tsinghua.edu.cn Received March 15, 2001; revised August 7, 2001. Abstract Automatic reconstruction of 3D ob jects from 2D orthographic views has b een a ma jor research issue in CAD/CAM. In this pap er, two accelerating techniques to improve the eÆciency of reconstruction are presented. First, some pseudo elements are removed by depth and top ology information as so on as the wire-frame is constructed, which reduces the searching space. Second, the prop osed algorithm do es not establish all p ossible surfaces in the pro cess of generating 3D faces. The surfaces and edge lo ops are generated by using the relationship b etween the b oundaries of 3D faces and their pro jections. This avoids the growth in combinational complexity of previous metho ds that have to check all p ossible pairs of 3D candidate edges. Keywords engineering drawing, wire-frame, reconstruction 1 Intro duction Reconstructing 3D ob jects from orthographic views has b een studied for the past two decades [1;2] . Algorithms attempting to solve automatic reconstruction can b e divided into CSG oriented and B-rep oriented approaches according to how they construct and represent the nal solid ob jects. The CSG ori