Labanotation handwriting recognition


Deutsch

LabanPad contains a handwriting recognition algorithm specialised in Labanotation. As the user writes down Labanotation symbols using a pen, they are analysed, tokenised and clearly displayed.

The Apple Newton is a pen based computer using a pressure sensitive display, therefore all input data is vector based. Using the same pen editing gestures are handled. The program has to decide if the input is a symbol or a gesture. For example if the user stays at the start of the drawing at the same spot for a recent time the program switches to the select mode.

Labanotation is mainly constructed of lines, points and circles, additionally half and quarter ellipses are used. First the input data is tested if the drawing is open or closed (all main direction symbols are closed) or if its a dot. Next the drawing is cut into single lines or part of ellipses. The position of every line and the relation inbetween them are analysed, filtered, sorted and crossing points calculated. The results are transferred by a fuzzy logic algorithm into the according token. After that the relation between the tokens are analysed and transferred to a semantic structure.


Project main page
My Homepage

Uni - Frankfurt Homepage


(C) 1996 Christian Griesbeck, Frankfurt/M
mail to: griesbec@stud.uni-frankfurt.de