Vector Quantisation Classifiers for Handwritten Character Recognition

Neschen, Martin (1995) Vector Quantisation Classifiers for Handwritten Character Recognition.
Published In: Parallel programming and applications : proceedings of the Workshop on Parallel Programming and Computations (ZEUS '95) and the 4th Nordic Transputer Conference (NTUG '95), Transputer and OCCAM engineering series. 45 IOS Press 1995, pp. 138-151.

Abstract

The development of a pattern recognition architecture based on vector quantization techniques is presented which is applied to the recognition of handwritten bank forms. After an overview of nearest-neighbor classification and clustering, a fast completely binary version of the k-means algorithm is introduced and results for latge character databases are given. An integration of these methods in a multi-agent environment is discussed. Both the efficient implementation on general MIMD processors and a realization on a dedicated SIMD architecture are presented.


Actions:
Download: [img] Postscript - Preprinted Version
Download (248kB) | Preview
Editorial actions: View Item View Item (Login required)
Content information:
Item Type: Proceedings article
Citations: 0 (Web of Science)
Uncontrolled Keywords:
Subjects:
Divisions: Mathematical Institute
Related URLs:
    Deposit Information:
    ZAIK Number: zpr95-185
    Depositing User: Archive Admin
    Date Deposited: 02 Apr 2001 00:00
    Last Modified: 19 Dec 2011 09:45
    URI: http://e-archive.informatik.uni-koeln.de/id/eprint/185