Visual programming paradigms for handwriting applications

The objective of the OSCAR (Offline Script and ChAracter Recognition toolset) project is to develop a prototyping environment for handwritten forms recognition applications which will allow a generic set of components to be rapidly configured to address any forms processing application. The toolkit consists of a visual programming framework together with components which include forms capture and layout specification, image preprocessing, feature extraction and classification algorithms for OCR, word recognition and signature verification, contextual postprocessing, and a unified set of evaluation databases. This paper describes the philosophy of the toolkit, design issues which are being addressed in building it, and some initial results which have been obtained using it