Project/Area Number |
03555072
|
Research Category |
Grant-in-Aid for Developmental Scientific Research (B)
|
Allocation Type | Single-year Grants |
Research Field |
情報工学
|
Research Institution | University of Tokyo |
Principal Investigator |
SAKAUCHI Masao University of Tokyo, Professor, 生産技術研究所, 教授 (30107370)
|
Co-Investigator(Kenkyū-buntansha) |
ZEN Heito University of Tokyo, Assistant Professor, 生産技術研究所, 助手 (20216568)
|
Project Period (FY) |
1991 – 1992
|
Project Status |
Completed (Fiscal Year 1992)
|
Budget Amount *help |
¥10,600,000 (Direct Cost: ¥10,600,000)
Fiscal Year 1992: ¥3,500,000 (Direct Cost: ¥3,500,000)
Fiscal Year 1991: ¥7,100,000 (Direct Cost: ¥7,100,000)
|
Keywords | Drawing Processing / Drawing Compiler / Model-based Vision / Database Capture / Map / Engineer Drawings / Spatial Data Structure / Rule Learning / 図面デ-タベ-ス / 図面認識 |
Research Abstract |
In this research project, effective environment for drawing understanding system realization, ie. drawing understanding system compiler, has been successfully developed. The following concrete results have been obtained. (1) Development multi-purpose drawing understanding system (a drawing compiler I) based on state-transition description of recognition rules: A state-transition type drawing compiler, where recognition level of objects in a drawing corresponds to a 'state' and bottom up and top down recognition rules correspond to 'transitions from state to another states' has been proposed and implemented by about a 6,500 lines Program of C and Prolog languages. Experiments for maps, hand-written drawings and relatively simple engineering drawings revealed the effectiveness of this system. (2) Development of a general purposed drawing complier II base on object-oriented model description of target drawings: Another type of drawing complier, where TMS and object-oriented model description of a target drawing are introduced, has been proposed and implemented by a about 5,000 line Prolog and 7,000 lines C program. Experiments for engineering drawings, layout diagrams revealed generality and effectiveness of the method. (3) Two suporting system also have successfully developed; one is original spatial data management structures and the other is a recognition rule learning system based on given positive and negative examples.
|