Pattern recognition and beautification for a pen based interface

The paper presents the algorithms for recognition and beautification which are used in incremental graphic design applications. These applications propose multimodal interfaces integrating handwriting, gesture, and speech on a pen-computer. User and computer collaborate to perform the task of incrementally designing a drawing. Processing and data representation take into account the variable quality of handwritten data, the man-machine interaction context and the cooperation between the user and the interpretation system. Both recognition algorithms may be used in combination in order to increase the speed and the set of recognized figures. Local recognition is followed by the beautification of the global structure in order to detect alignments and logical structures. The beautification enables the user to display a clean version of the original draft. The applications which the authors developed are used to recognize tables, gestures, geometrical figures or diagram networks