Feature enhancement for extracting on-line isolated handwritten characters

The study of online handwriting recognition has gained an immense interest among the researchers especially with the increase in use of the personal digital assistant (PDA). The large number of writing styles and the variability between them make the handwriting recognition a challenging area to date. The present tools for modelling are not sufficient to cater for the various styles of human handwriting. Furthermore, the techniques used to get appropriate features, architecture and network parameters for online handwriting recognition are still ineffective. The success of any recognition system depends critically upon how far a set of appropriate numerical attributes or features can be extracted from the object of interest for the purpose of recognition. Thus the aim of this research work is to propose novel feature extraction methods to facilitate a system or device to achieve satisfactory online handwriting recognition. Two new simple and robust methods based on annotated image and sub-character primitive feature extractions have been proposed. The selection of features is based mainly on their effectiveness. Using the proposed techniques and a neural network based classifier, several experiments were carried out using the UNIPEN benchmark database. The techniques are independent of character size and can extract features from raw data without resizing. The maximum recognition rates achieved are 94% and 92% for annotated image and subcharacter primitive methods respectively