HANDWRITING RECOGNITION ACCURACY VERSUS TABLET RESOLUTION AND SAMPLING RATE.

Abstract

In the development of an on-line handwriting recognition system, it is desirable to establish a balance between the input device characteristics and the recognition algorithm. Using a high performance tablet, writing samples of discretely written English alphabetic and numeric characters were collected and modified by software simulation into various resolution and sampling rate combinations. They were then processed by an elastic match recognition algorithm to determine the effect on recognition accuracy. Preliminary results suggest that there is an optimal set of tablet parameters. Future study will involve more writing samples and writers.