Fiber bundle image restoration using deep learning

Jianbo Shao, Junchao Zhang, Xiao Huang, Rongguang Liang, Jacobus J Barnard

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

We propose a deep learning-based restoration method to remove honeycomb patterns and improve resolution for fiber bundle (FB) images. By building and calibrating a dual-sensor imaging system, we capture FB images and corresponding ground truth data to train the network. Images without fiber bundle fixed patterns are restored from raw FB images as direct inputs, and spatial resolution is significantly enhanced for the trained sample type. We also construct the brightness mapping between the two image types for the effective use of all data, providing the ability to output images of the expected brightness. We evaluate our framework with data obtained from lens tissues and human histological specimens using both objective and subjective measures.

Original languageEnglish (US)
Pages (from-to)1080-1083
Number of pages4
JournalOptics letters
Volume44
Issue number5
DOIs
StatePublished - Mar 1 2019

Fingerprint

restoration
learning
bundles
fibers
brightness
ground truth
calibrating
spatial resolution
lenses
output
sensors

ASJC Scopus subject areas

  • Atomic and Molecular Physics, and Optics

Cite this

Fiber bundle image restoration using deep learning. / Shao, Jianbo; Zhang, Junchao; Huang, Xiao; Liang, Rongguang; Barnard, Jacobus J.

In: Optics letters, Vol. 44, No. 5, 01.03.2019, p. 1080-1083.

Research output: Contribution to journalArticle

Shao, Jianbo ; Zhang, Junchao ; Huang, Xiao ; Liang, Rongguang ; Barnard, Jacobus J. / Fiber bundle image restoration using deep learning. In: Optics letters. 2019 ; Vol. 44, No. 5. pp. 1080-1083.
@article{5f98d69d48ce4521bdd70879a7be3ce4,
title = "Fiber bundle image restoration using deep learning",
abstract = "We propose a deep learning-based restoration method to remove honeycomb patterns and improve resolution for fiber bundle (FB) images. By building and calibrating a dual-sensor imaging system, we capture FB images and corresponding ground truth data to train the network. Images without fiber bundle fixed patterns are restored from raw FB images as direct inputs, and spatial resolution is significantly enhanced for the trained sample type. We also construct the brightness mapping between the two image types for the effective use of all data, providing the ability to output images of the expected brightness. We evaluate our framework with data obtained from lens tissues and human histological specimens using both objective and subjective measures.",
author = "Jianbo Shao and Junchao Zhang and Xiao Huang and Rongguang Liang and Barnard, {Jacobus J}",
year = "2019",
month = "3",
day = "1",
doi = "10.1364/OL.44.001080",
language = "English (US)",
volume = "44",
pages = "1080--1083",
journal = "Optics Letters",
issn = "0146-9592",
publisher = "The Optical Society",
number = "5",

}

TY - JOUR

T1 - Fiber bundle image restoration using deep learning

AU - Shao, Jianbo

AU - Zhang, Junchao

AU - Huang, Xiao

AU - Liang, Rongguang

AU - Barnard, Jacobus J

PY - 2019/3/1

Y1 - 2019/3/1

N2 - We propose a deep learning-based restoration method to remove honeycomb patterns and improve resolution for fiber bundle (FB) images. By building and calibrating a dual-sensor imaging system, we capture FB images and corresponding ground truth data to train the network. Images without fiber bundle fixed patterns are restored from raw FB images as direct inputs, and spatial resolution is significantly enhanced for the trained sample type. We also construct the brightness mapping between the two image types for the effective use of all data, providing the ability to output images of the expected brightness. We evaluate our framework with data obtained from lens tissues and human histological specimens using both objective and subjective measures.

AB - We propose a deep learning-based restoration method to remove honeycomb patterns and improve resolution for fiber bundle (FB) images. By building and calibrating a dual-sensor imaging system, we capture FB images and corresponding ground truth data to train the network. Images without fiber bundle fixed patterns are restored from raw FB images as direct inputs, and spatial resolution is significantly enhanced for the trained sample type. We also construct the brightness mapping between the two image types for the effective use of all data, providing the ability to output images of the expected brightness. We evaluate our framework with data obtained from lens tissues and human histological specimens using both objective and subjective measures.

UR - http://www.scopus.com/inward/record.url?scp=85062227168&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85062227168&partnerID=8YFLogxK

U2 - 10.1364/OL.44.001080

DO - 10.1364/OL.44.001080

M3 - Article

C2 - 30821775

AN - SCOPUS:85062227168

VL - 44

SP - 1080

EP - 1083

JO - Optics Letters

JF - Optics Letters

SN - 0146-9592

IS - 5

ER -