Advances in Handwriting Recognition

Abstract

On-line handwriting recognition by discrete HMM with fast learning diacritical processing using efficient accounting procedures in a forward search a handwritten form reader architecture combining different classifiers and levels of knowledge - a first step towards an adaptive recognition system architecture for handwritten text recognition systems search algorithms for the recognition of cursive phrases without world segmentation a method for the determination of features used in human reading of cursive handwriting global methods for stroke segmentation an advanced segmentation technique for cursive word recognition document understanding based on maximum a posteriori probability estimation combining shape matrices and HMMs for hand-drawn pictogram recognition.