• DocumentCode
    1138395
  • Title

    Computer Description of Bodies Bounded by Quadric Surfaces from a Set of Imperfect Projections

  • Author

    Shapira, Ruth ; Freeman, Herbert

  • Author_Institution
    Rensselaer Polytechnic Institute
  • Issue
    9
  • fYear
    1978
  • Firstpage
    841
  • Lastpage
    854
  • Abstract
    This paper describes a computer program for constructing a description of solid bodies from a set of pictures taken from different vantage points. The bodies are assumed to be bounded by faces which are planar or quadric, and to have vertices formed by exactly three faces It is assumed that a preprocessor provides the program with line and junction information which it has extracted from the pictures. The preprocessor is expected to make mistakes, such as losing features or providing misinformation about their nature. A technique is presented for validating doubtful features as well as for matching corresponding features extracted from the different pictures. New grammar rules are developed for linedrawing projections of curved and planar bodies and are used as a tool in the scene analysis process. Each picture´s data analysis is supported dynamically by the results obtained thus far in the other pictures´ analysis. The analyzed data from all pictures are grouped into sets, each corresponding to a single face (flat or curved), whose nature is also determined. The sets are grouped again to correspond to the different bodies in the scene. The program written in PL/I has been tested successfully on several scenes.
  • Keywords
    Artificial intelligence; computer graphics; pattern recognition; picture grammars; scene analysis; three-dimensional reconstruction; Computer graphics; Data analysis; Data mining; Feature extraction; Image analysis; Layout; Pattern recognition; Solids; Testing; Artificial intelligence; computer graphics; pattern recognition; picture grammars; scene analysis; three-dimensional reconstruction;
  • fLanguage
    English
  • Journal_Title
    Computers, IEEE Transactions on
  • Publisher
    ieee
  • ISSN
    0018-9340
  • Type

    jour

  • DOI
    10.1109/TC.1978.1675204
  • Filename
    1675204