A Neuromorphic Architecture for Context Aware Text Image Recognition

Qinru Qiu, Zhe Li, Khadeer Ahmed, Wei Liu, Syed Faisal Habib, Hai (Helen) Li, Miao Hu

Research output: Contribution to journalArticle

4 Citations (Scopus)

Abstract

Although existing optical character recognition (OCR) tools can achieve excellent performance in text image detection and pattern recognition, they usually require a clean input image. Most of them do not perform well when the image is partially occluded or smudged. Humans are able to tolerate much worse image quality during reading because the perception errors can be corrected by the knowledge in word and sentence level context. In this paper, we present a brain-inspired information processing framework for context-aware Intelligent Text Recognition (ITR) and its acceleration using memristor based crossbar array. The ITRS has a bottom layer of massive parallel Brain-state-in-a-box (BSB) engines that give fuzzy pattern matching results and an upper layer of statistical inference based error correction. Optimizations on each layer of the framework are introduced to improve system performance. A parallel architecture is presented that incorporates the memristor crossbar array to accelerate the pattern matching. Compared to traditional multicore microprocessor, the accelerator has the potential to provide tremendous area and power savings and more than 8,000 times speedups.

Original languageEnglish (US)
JournalJournal of Signal Processing Systems
DOIs
StateAccepted/In press - Nov 4 2015

Fingerprint

Memristors
Image recognition
Image Recognition
Pattern matching
Context-aware
Brain
Pattern Matching
Optical character recognition
Parallel architectures
Error correction
Image quality
Pattern recognition
Particle accelerators
Microprocessor chips
Power Saving
Character Recognition
Parallel Architectures
Microprocessor
Error Correction
Accelerator

Keywords

  • Memristor crossbar array
  • Neuromorphic
  • Text recognition

ASJC Scopus subject areas

  • Control and Systems Engineering
  • Modeling and Simulation
  • Hardware and Architecture
  • Information Systems
  • Signal Processing
  • Theoretical Computer Science

Cite this

A Neuromorphic Architecture for Context Aware Text Image Recognition. / Qiu, Qinru; Li, Zhe; Ahmed, Khadeer; Liu, Wei; Habib, Syed Faisal; Li, Hai (Helen); Hu, Miao.

In: Journal of Signal Processing Systems, 04.11.2015.

Research output: Contribution to journalArticle

Qiu, Qinru ; Li, Zhe ; Ahmed, Khadeer ; Liu, Wei ; Habib, Syed Faisal ; Li, Hai (Helen) ; Hu, Miao. / A Neuromorphic Architecture for Context Aware Text Image Recognition. In: Journal of Signal Processing Systems. 2015.
@article{662c04a7bf014e2488fb8e8723b42a69,
title = "A Neuromorphic Architecture for Context Aware Text Image Recognition",
abstract = "Although existing optical character recognition (OCR) tools can achieve excellent performance in text image detection and pattern recognition, they usually require a clean input image. Most of them do not perform well when the image is partially occluded or smudged. Humans are able to tolerate much worse image quality during reading because the perception errors can be corrected by the knowledge in word and sentence level context. In this paper, we present a brain-inspired information processing framework for context-aware Intelligent Text Recognition (ITR) and its acceleration using memristor based crossbar array. The ITRS has a bottom layer of massive parallel Brain-state-in-a-box (BSB) engines that give fuzzy pattern matching results and an upper layer of statistical inference based error correction. Optimizations on each layer of the framework are introduced to improve system performance. A parallel architecture is presented that incorporates the memristor crossbar array to accelerate the pattern matching. Compared to traditional multicore microprocessor, the accelerator has the potential to provide tremendous area and power savings and more than 8,000 times speedups.",
keywords = "Memristor crossbar array, Neuromorphic, Text recognition",
author = "Qinru Qiu and Zhe Li and Khadeer Ahmed and Wei Liu and Habib, {Syed Faisal} and Li, {Hai (Helen)} and Miao Hu",
year = "2015",
month = "11",
day = "4",
doi = "10.1007/s11265-015-1067-4",
language = "English (US)",
journal = "Journal of Signal Processing Systems",
issn = "1939-8018",
publisher = "Springer New York",

}

TY - JOUR

T1 - A Neuromorphic Architecture for Context Aware Text Image Recognition

AU - Qiu, Qinru

AU - Li, Zhe

AU - Ahmed, Khadeer

AU - Liu, Wei

AU - Habib, Syed Faisal

AU - Li, Hai (Helen)

AU - Hu, Miao

PY - 2015/11/4

Y1 - 2015/11/4

N2 - Although existing optical character recognition (OCR) tools can achieve excellent performance in text image detection and pattern recognition, they usually require a clean input image. Most of them do not perform well when the image is partially occluded or smudged. Humans are able to tolerate much worse image quality during reading because the perception errors can be corrected by the knowledge in word and sentence level context. In this paper, we present a brain-inspired information processing framework for context-aware Intelligent Text Recognition (ITR) and its acceleration using memristor based crossbar array. The ITRS has a bottom layer of massive parallel Brain-state-in-a-box (BSB) engines that give fuzzy pattern matching results and an upper layer of statistical inference based error correction. Optimizations on each layer of the framework are introduced to improve system performance. A parallel architecture is presented that incorporates the memristor crossbar array to accelerate the pattern matching. Compared to traditional multicore microprocessor, the accelerator has the potential to provide tremendous area and power savings and more than 8,000 times speedups.

AB - Although existing optical character recognition (OCR) tools can achieve excellent performance in text image detection and pattern recognition, they usually require a clean input image. Most of them do not perform well when the image is partially occluded or smudged. Humans are able to tolerate much worse image quality during reading because the perception errors can be corrected by the knowledge in word and sentence level context. In this paper, we present a brain-inspired information processing framework for context-aware Intelligent Text Recognition (ITR) and its acceleration using memristor based crossbar array. The ITRS has a bottom layer of massive parallel Brain-state-in-a-box (BSB) engines that give fuzzy pattern matching results and an upper layer of statistical inference based error correction. Optimizations on each layer of the framework are introduced to improve system performance. A parallel architecture is presented that incorporates the memristor crossbar array to accelerate the pattern matching. Compared to traditional multicore microprocessor, the accelerator has the potential to provide tremendous area and power savings and more than 8,000 times speedups.

KW - Memristor crossbar array

KW - Neuromorphic

KW - Text recognition

UR - http://www.scopus.com/inward/record.url?scp=84946104801&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84946104801&partnerID=8YFLogxK

U2 - 10.1007/s11265-015-1067-4

DO - 10.1007/s11265-015-1067-4

M3 - Article

AN - SCOPUS:84946104801

JO - Journal of Signal Processing Systems

JF - Journal of Signal Processing Systems

SN - 1939-8018

ER -