B. Zhu, T. Shimamura, and M. Nakagawa (Japan)
Form processing, Form recognition, Labeling,Morphology, Paper-based UI
This paper describes document image processing methods for active forms that employ dot texture for printing input frames in the forms and encode how to process themselves in the dot texture. The dot texture is the texture made up of small dots. In the domain of form processing, separation of filled-in handwritings from frames has been studied. By employing the dot texture, handwritings are easily separated from frames under monochrome printing/reading environments even if they overlap on the frames. Moreover, by embedding the information about attributes of handwritings and how to process them in the dot texture, the recognition rate of handwritings is improved, and the form itself directs the form processing rather than the form reading machine. Thus, the reading machine is a general-purpose machine and various forms inputted into a single reader are distinctly processed as specified by each form. This paper also presents the preliminary evaluation of the described methods.
Important Links:
Go Back
