A computerized transcription system for cursive shorthand writing

A structural recognition approach to the problem of computerized transcription of cursive shorthand (SH) writing is described. Methods of extracting critical points and segmentations of SH outlines are presented. Critical points are cusps, self-intersections, and corners (discontinuities in curvature). The authors have designed a software that recognizes SH words written separately and made up of SH characters separated by permutations of critical points. The results show that computerized transcription of cursive SH is a practical possibility.<
>