Paper Details

ICR: A Transputer-Based Intelligent Character Reader

Authors: db_connect: Could not connect to paper db at "wotug@dragon.kent.ac.uk"
db_connect: Could not connect to paper db at "wotug@dragon.kent.ac.uk"
Wong F.S., Francis, Seng, Koh Liang

Abstract:

This paper presents the design of a transputer-based Intelligent Character Reader (ICR). The implementation of the ICR is based on two algorithms. The Filtered Projection Technique (FPT) is a modification of the conventional projection profile techniques - it filters out and projects the structural information of a visual pattern into several components along several directions, such that some salient structural information can be extracted from these components easily. The Structural Complexity Index (SCI) algorithm determines the structural complexity of a visual pattern and produces the corresponding index to indicate its relative complexity. The ICR is capable of recognizing printed characters, such as the Chinese characters printedon the local newspapers, accurately, despite the presence of noise due to printing, scanning, slight font variations and misalignment. The ICR prototype is implemented on Occam II and C, and run on an array of transputers.

Proceedings:

OUG-11: Developing Transputer Applications, db_connect: Could not connect to paper db at "wotug@dragon.kent.ac.uk"
J. Wexler, 1989, pp 182 - 189 db_connect: Could not connect to paper db at "wotug@dragon.kent.ac.uk"
published by IOS Press, Amsterdam

This record in other formats: