Recognition of Cursive Writing on Personal Checks
Andr As Kornai, Khalid Mohiuddin, Scott D. Connell · SZTAKI Publication Repository (Hungarian Academy of Sciences) · 1996
The system described in this paper applies Hidden Markov technology both to the task of recognizing the cursive legal amount on personal checks and the isolated (numeric) courtesy amount. 0. INTRODUCTION In this paper we describe the main components of a bank check amount recognition system prototype currently under development at the IBM Almaden Research Center. The goal of the system is to enhance the performance of a commercially available check recognition system by performing recognition not only on the isolated (numeric) courtesy amount but also on the cursive handwritten legal amount that appears on standard US personal checks. In addition to the core handwriting recognition task, such a system presents an additional challenge, the reliable segmentation of the image into recognizable units. Section 1 presents the principal preprocessing steps including deskewing, zone finding, and feature extraction. Section 2 describes the combined segmentation and recognition module, which ...