Title :
On-line training of freehand curve identifier FSCI in sketch-based CAD system
Author :
Doushita, Takahiro ; Saga, Sato
Author_Institution :
Dept. of Comput. Sci. & Syst. Eng., Muroran Inst. of Technol., Muroran
Abstract :
As a freehand curve recognition method for sketch-based CAD systems, we previously proposed the Fuzzy Spline Curve Identifier (FSCI), which identifies a freehand curve as one of seven types of primitive geometric curve based on the drawing style of the user and the curve shape. Moreover, we have realized a trainable version of FSCI in which the identification characteristics can be tuned according to each user´s drawing data by the back-propagation learning algorithm. However, users are required to draw a considerable number of primitive geometric curves to collect enough drawing data to tune the FSCI before starting practical drawing on a sketch-based CAD system. The present paper proposes an on-line training mechanism for FSCI in which its identification characteristics are adaptively and automatically tuned while a user is operating the sketch-based CAD system.
Keywords :
CAD; computer aided instruction; fuzzy set theory; neural nets; splines (mathematics); CAD system; back-propagation learning algorithm; freehand curve identifier; fuzzy spline curve identifier; on-line training; Application software; Computer applications; Computer industry; Fuzzy neural networks; Fuzzy systems; Industrial training; Power capacitors; Shape; Spline; Usability;
Conference_Titel :
Soft Computing in Industrial Applications, 2008. SMCia '08. IEEE Conference on
Conference_Location :
Muroran
Print_ISBN :
978-1-4244-3782-5
Electronic_ISBN :
978-4-9904-2590-6
DOI :
10.1109/SMCIA.2008.5045952