A modular neural network classifier for the recognition of occluded characters in automatic license plate reading

JAG Nijhuis; A Broersma; L Spaanenburg

A modular neural network classifier for the recognition of occluded characters in automatic license plate reading

JAG Nijhuis^*
, A Broersma
, L Spaanenburg

^*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

Abstract

Occlusion is the most common reason for lowered recognition yield in free-flow license-plate reading systems. (Non-)occluded characters can readily be learned in separate neural networks but not together. Even a small proportion of occluded characters in the training set will already significantly reduce the overall recognition yield. This paper shows that a modular network can handle a realistic mixture of (non-) occluded characters with a 99.8% recognition yield per character.

Original language	English
Title of host publication	COMPUTATIONAL INTELLIGENT SYSTEMS FOR APPLIED RESEARCH
Editors	D Ruan, P Dhondt, EE Kerre
Place of Publication	SINGAPORE
Publisher	World Scientific Publishing
Pages	363-372
Number of pages	10
ISBN (Print)	981-238-066-3
Publication status	Published - 2002
Event	5th International Conference on Fuzzy Logic and Intelligent Technologies in Nuclear Science - , Belgium Duration: 16-Sept-2002 → 18-Sept-2002

Other

Other	5th International Conference on Fuzzy Logic and Intelligent Technologies in Nuclear Science
Country/Territory	Belgium
Period	16/09/2002 → 18/09/2002

Keywords

MULTIPLE CLASSIFIERS

Handle.net

Persistent link

Cite this

@inproceedings{12c0c8dcafbc477489f385a8fbe19686,

title = "A modular neural network classifier for the recognition of occluded characters in automatic license plate reading",

abstract = "Occlusion is the most common reason for lowered recognition yield in free-flow license-plate reading systems. (Non-)occluded characters can readily be learned in separate neural networks but not together. Even a small proportion of occluded characters in the training set will already significantly reduce the overall recognition yield. This paper shows that a modular network can handle a realistic mixture of (non-) occluded characters with a 99.8\% recognition yield per character.",

keywords = "MULTIPLE CLASSIFIERS",

author = "JAG Nijhuis and A Broersma and L Spaanenburg",

year = "2002",

language = "English",

isbn = "981-238-066-3",

pages = "363--372",

editor = "D Ruan and P Dhondt and EE Kerre",

booktitle = "COMPUTATIONAL INTELLIGENT SYSTEMS FOR APPLIED RESEARCH",

publisher = "World Scientific Publishing",

note = "5th International Conference on Fuzzy Logic and Intelligent Technologies in Nuclear Science ; Conference date: 16-09-2002 Through 18-09-2002",

}

Nijhuis, JAG, Broersma, A & Spaanenburg, L 2002, A modular neural network classifier for the recognition of occluded characters in automatic license plate reading. in D Ruan, P Dhondt & EE Kerre (eds), COMPUTATIONAL INTELLIGENT SYSTEMS FOR APPLIED RESEARCH. World Scientific Publishing, SINGAPORE, pp. 363-372, 5th International Conference on Fuzzy Logic and Intelligent Technologies in Nuclear Science, Belgium, 16/09/2002.

A modular neural network classifier for the recognition of occluded characters in automatic license plate reading. / Nijhuis, JAG; Broersma, A; Spaanenburg, L.
COMPUTATIONAL INTELLIGENT SYSTEMS FOR APPLIED RESEARCH. ed. / D Ruan; P Dhondt; EE Kerre. SINGAPORE: World Scientific Publishing, 2002. p. 363-372.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

TY - GEN

T1 - A modular neural network classifier for the recognition of occluded characters in automatic license plate reading

AU - Nijhuis, JAG

AU - Broersma, A

AU - Spaanenburg, L

PY - 2002

Y1 - 2002

N2 - Occlusion is the most common reason for lowered recognition yield in free-flow license-plate reading systems. (Non-)occluded characters can readily be learned in separate neural networks but not together. Even a small proportion of occluded characters in the training set will already significantly reduce the overall recognition yield. This paper shows that a modular network can handle a realistic mixture of (non-) occluded characters with a 99.8% recognition yield per character.

AB - Occlusion is the most common reason for lowered recognition yield in free-flow license-plate reading systems. (Non-)occluded characters can readily be learned in separate neural networks but not together. Even a small proportion of occluded characters in the training set will already significantly reduce the overall recognition yield. This paper shows that a modular network can handle a realistic mixture of (non-) occluded characters with a 99.8% recognition yield per character.

KW - MULTIPLE CLASSIFIERS

M3 - Conference contribution

SN - 981-238-066-3

SP - 363

EP - 372

BT - COMPUTATIONAL INTELLIGENT SYSTEMS FOR APPLIED RESEARCH

A2 - Ruan, D

A2 - Dhondt, P

A2 - Kerre, EE

PB - World Scientific Publishing

CY - SINGAPORE

T2 - 5th International Conference on Fuzzy Logic and Intelligent Technologies in Nuclear Science

Y2 - 16 September 2002 through 18 September 2002

ER -

A modular neural network classifier for the recognition of occluded characters in automatic license plate reading

Abstract

Other

Keywords

Handle.net

Fingerprint

Cite this