|
Abstract : |
Cursive script recognition has historically received relatively little attention until recently. In contrast, there has been tremendous research effort in the speech recognition area for a long time. Much of what has been learned in speech recognition deals with language and stochastic modeling techniques that can be applied to the on-line handwriting recognition problem almost directly. This paper describes preliminary results from the application of sophisticated speech language/stochastic modeling methods to the online handwriting recognition problem. Early results with simple features are very encouraging. This effort has also helped in developing a generalized language modeling system that can be used for both speech and handwriting recognition. 1., |
