Efficient Image Preprocessing of Digital Holograms of Marine Plankton

Zonghua Liu*, John Watson, Alastair Allen

*Corresponding author for this work

Research output: Contribution to journalReview article

4 Citations (Scopus)

Abstract

A set of image preprocessing approaches are developed for processing plankton images reconstructed from digital holograms. First, a threshold-based algorithm of image segmentation is proposed and applied to extract the regions of plankton from the original digital images. To improve the performance of image segmentation, an appropriate filter is adopted to reduce the background noise from the image and the image gray level is adjusted to enhance the image contrast. Second, we develop a novel and efficient edge detection method purposefully for the binary images. Third, we propose and use a simple chain-code-based algorithm to eliminate the single-pixel branches along the shape boundary, which will help boundary tracing work stably. Then, an algorithm is improved and applied to trace the boundaries of the plankton regions. This algorithm is optimized based on the relationship between two consecutive chain-codes such that it is fast on implementation. Finally, break points of the shape boundary are efficiently detected based on chain-codes and the boundary is represented compactly by a polygon comprised of those points. After images are preprocessed by these approaches, some redundant information of shape is reduced that will accelerate the running speeds of further image processing and aid identification and classification of plankton at species level. We analyze the accuracy and efficiency of our algorithms. The results show that our algorithm of image segmentation has a good performance in accuracy. Our edge detection method also outperforms the commonly used edge detection methods in terms of localization performance and the running time.

Original languageEnglish
Pages (from-to)83-92
Number of pages10
JournalIEEE Journal of Oceanic Engineering
Volume43
Issue number1
Early online date24 Apr 2017
DOIs
Publication statusPublished - Jan 2018

Fingerprint

Plankton
Holograms
Edge detection
Image segmentation
Binary images
Image processing
Pixels
Processing

Keywords

  • Digital holography
  • image processing
  • image segmentation
  • marine plankton
  • shape polygonal description

ASJC Scopus subject areas

  • Ocean Engineering
  • Mechanical Engineering
  • Electrical and Electronic Engineering

Cite this

Efficient Image Preprocessing of Digital Holograms of Marine Plankton. / Liu, Zonghua; Watson, John; Allen, Alastair.

In: IEEE Journal of Oceanic Engineering, Vol. 43, No. 1, 01.2018, p. 83-92.

Research output: Contribution to journalReview article

@article{b5d86906cdf34e60854121a9558cc3dd,
title = "Efficient Image Preprocessing of Digital Holograms of Marine Plankton",
abstract = "A set of image preprocessing approaches are developed for processing plankton images reconstructed from digital holograms. First, a threshold-based algorithm of image segmentation is proposed and applied to extract the regions of plankton from the original digital images. To improve the performance of image segmentation, an appropriate filter is adopted to reduce the background noise from the image and the image gray level is adjusted to enhance the image contrast. Second, we develop a novel and efficient edge detection method purposefully for the binary images. Third, we propose and use a simple chain-code-based algorithm to eliminate the single-pixel branches along the shape boundary, which will help boundary tracing work stably. Then, an algorithm is improved and applied to trace the boundaries of the plankton regions. This algorithm is optimized based on the relationship between two consecutive chain-codes such that it is fast on implementation. Finally, break points of the shape boundary are efficiently detected based on chain-codes and the boundary is represented compactly by a polygon comprised of those points. After images are preprocessed by these approaches, some redundant information of shape is reduced that will accelerate the running speeds of further image processing and aid identification and classification of plankton at species level. We analyze the accuracy and efficiency of our algorithms. The results show that our algorithm of image segmentation has a good performance in accuracy. Our edge detection method also outperforms the commonly used edge detection methods in terms of localization performance and the running time.",
keywords = "Digital holography, image processing, image segmentation, marine plankton, shape polygonal description",
author = "Zonghua Liu and John Watson and Alastair Allen",
year = "2018",
month = "1",
doi = "10.1109/JOE.2017.2690537",
language = "English",
volume = "43",
pages = "83--92",
journal = "IEEE Journal of Oceanic Engineering",
issn = "0364-9059",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "1",

}

TY - JOUR

T1 - Efficient Image Preprocessing of Digital Holograms of Marine Plankton

AU - Liu, Zonghua

AU - Watson, John

AU - Allen, Alastair

PY - 2018/1

Y1 - 2018/1

N2 - A set of image preprocessing approaches are developed for processing plankton images reconstructed from digital holograms. First, a threshold-based algorithm of image segmentation is proposed and applied to extract the regions of plankton from the original digital images. To improve the performance of image segmentation, an appropriate filter is adopted to reduce the background noise from the image and the image gray level is adjusted to enhance the image contrast. Second, we develop a novel and efficient edge detection method purposefully for the binary images. Third, we propose and use a simple chain-code-based algorithm to eliminate the single-pixel branches along the shape boundary, which will help boundary tracing work stably. Then, an algorithm is improved and applied to trace the boundaries of the plankton regions. This algorithm is optimized based on the relationship between two consecutive chain-codes such that it is fast on implementation. Finally, break points of the shape boundary are efficiently detected based on chain-codes and the boundary is represented compactly by a polygon comprised of those points. After images are preprocessed by these approaches, some redundant information of shape is reduced that will accelerate the running speeds of further image processing and aid identification and classification of plankton at species level. We analyze the accuracy and efficiency of our algorithms. The results show that our algorithm of image segmentation has a good performance in accuracy. Our edge detection method also outperforms the commonly used edge detection methods in terms of localization performance and the running time.

AB - A set of image preprocessing approaches are developed for processing plankton images reconstructed from digital holograms. First, a threshold-based algorithm of image segmentation is proposed and applied to extract the regions of plankton from the original digital images. To improve the performance of image segmentation, an appropriate filter is adopted to reduce the background noise from the image and the image gray level is adjusted to enhance the image contrast. Second, we develop a novel and efficient edge detection method purposefully for the binary images. Third, we propose and use a simple chain-code-based algorithm to eliminate the single-pixel branches along the shape boundary, which will help boundary tracing work stably. Then, an algorithm is improved and applied to trace the boundaries of the plankton regions. This algorithm is optimized based on the relationship between two consecutive chain-codes such that it is fast on implementation. Finally, break points of the shape boundary are efficiently detected based on chain-codes and the boundary is represented compactly by a polygon comprised of those points. After images are preprocessed by these approaches, some redundant information of shape is reduced that will accelerate the running speeds of further image processing and aid identification and classification of plankton at species level. We analyze the accuracy and efficiency of our algorithms. The results show that our algorithm of image segmentation has a good performance in accuracy. Our edge detection method also outperforms the commonly used edge detection methods in terms of localization performance and the running time.

KW - Digital holography

KW - image processing

KW - image segmentation

KW - marine plankton

KW - shape polygonal description

UR - http://www.scopus.com/inward/record.url?scp=85018959419&partnerID=8YFLogxK

U2 - 10.1109/JOE.2017.2690537

DO - 10.1109/JOE.2017.2690537

M3 - Review article

VL - 43

SP - 83

EP - 92

JO - IEEE Journal of Oceanic Engineering

JF - IEEE Journal of Oceanic Engineering

SN - 0364-9059

IS - 1

ER -