• DocumentCode
    790507
  • Title

    CSG-EESI: a new solid representation scheme and a conversion expert system

  • Author

    Xie, Shun-En ; Calvert, Thomas W.

  • Author_Institution
    Sch. of Comput. Sci., Simon Fraser Univ., Burnaby, BC, Canada
  • Volume
    10
  • Issue
    2
  • fYear
    1988
  • fDate
    3/1/1988 12:00:00 AM
  • Firstpage
    221
  • Lastpage
    234
  • Abstract
    A representation scheme is proposed for describing 3-D mechanical parts and structural bodies which are formed from planes and quadratic faces. Since the method combines features of the constructive solid geometry (CSG) representation and an extension of the enhanced (EESI) spherical image representation (ESI), it is designated the CSG-EESI. In this scheme, the body model is roughly divided into two levels: the higher level corresponds to a restricted CSG tree that contains the structural information describing how the various subparts form the body; the lower level contains the geometric information for those simple subparts and represents them by an extension of enhanced spherical images. The scheme can be used both as the medium between pictorial models and relational models and as an internal model to facilitate the recognition of bodies. An expert system written in C-PROLOG on a VAX 11/750 is presented that converts BR-like models into the CSG-EESI representation is presented
  • Keywords
    computer vision; computerised pattern recognition; expert systems; trees (mathematics); 3D solid object representation; C-PROLOG; CSG-EESI; VAX 11/750; computer vision; constructive solid geometry; enhanced spherical image representation; expert system; geometric information; pictorial models; relational models; Computer graphics; Computer vision; Design methodology; Expert systems; Face detection; Image converters; Image representation; Information geometry; Machine vision; Solid modeling;
  • fLanguage
    English
  • Journal_Title
    Pattern Analysis and Machine Intelligence, IEEE Transactions on
  • Publisher
    ieee
  • ISSN
    0162-8828
  • Type

    jour

  • DOI
    10.1109/34.3884
  • Filename
    3884