Visual Learning and Recognition by Computer.
TEXAS UNIV AUSTIN ELECTRONICS RESEARCH CENTER
Pagination or Media Count:
The objective of the study is to formulate an algorithm for forming a description of three-dimensional objects which will be invariant with the objects position in three-space. The objects are three-dimensional line drawing description of the object by edge following techniques. The object is rotated in space so that all surfaces of the object are observed and the interconnections of all surfaces are determined. The line drawing descriptions of the object are used by the computer to form a complete description of the object which is position invariant. No previous knowledge about the object being observed is supplied to the computer program except that the surfaces of the object are planar. After a library of different objects has been obtained by the computer, any one of these objects may be placed before the television camera in any position. The computer will recognize and name the object being observed by matching its description to a description stored in the library. Author