Unsupervised segmentation of underwater objects in sonar images

Avi Abu; Roee Diamant

doi:10.1109/OCEANSE.2017.8084853

Unsupervised segmentation of underwater objects in sonar images

Avi Abu, Roee Diamant

Department of Marine Technologies

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

Abstract

We focus on the segmentation of sonar images for the aim of underwater object detection. Speckle noise and intensity inhomogeneity may cause false detections, and complex seabed textures like sand-ripples and sea-grass often lead to false segmentation. To tackle these problems, we propose a new method to incorporate the possible spatial correlation between neighboring pixels in the sonar image for improved segmentation. Our method modifies the expectation-maximization (EM) algorithm by adding an intermediate step (I-step) between the expectation (E-step) and maximization (M-step). Results show that our proposed method achieves improved segmentation performance over the state-of-the-art and is also robust to different seabed texture and for intensity inhomogeneity.

Original language	English
Title of host publication	OCEANS 2017 - Aberdeen
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	1-5
Number of pages	5
ISBN (Electronic)	9781509052783
DOIs	https://doi.org/10.1109/OCEANSE.2017.8084853
State	Published - 25 Oct 2017
Event	OCEANS 2017 - Aberdeen - Aberdeen, United Kingdom Duration: 19 Jun 2017 → 22 Jun 2017

Publication series

Name	OCEANS 2017 - Aberdeen
Volume	2017-October

Conference

Conference	OCEANS 2017 - Aberdeen
Country/Territory	United Kingdom
City	Aberdeen
Period	19/06/17 → 22/06/17

Bibliographical note

Funding Information:
Our second recommendation relates to financial support of students/trainees, from the undergraduate to postdoctoral levels. Programs should set aside training and travel monies for ethnic minority students/trainees. Such funds could cover tuition, stipends, and professional development to allow for students/trainees to attend national conferences, since these students may derive particular benefit from connecting with ethnic minority students in other programs and from connecting with a nationwide network of ethnic minority neuropsychologists present at these meetings.

Publisher Copyright:
© 2017 IEEE.

Keywords

Expectation-maximization (EM)
gamma distribution
object detection
sand ripples
sonar image segmentation
speckle noise

ASJC Scopus subject areas

Instrumentation
Computer Networks and Communications
Oceanography
Acoustics and Ultrasonics
Automotive Engineering

Access to Document

10.1109/OCEANSE.2017.8084853

Cite this

@inproceedings{bf0388675ab84adb9ca6d1ee742f7895,

title = "Unsupervised segmentation of underwater objects in sonar images",

abstract = "We focus on the segmentation of sonar images for the aim of underwater object detection. Speckle noise and intensity inhomogeneity may cause false detections, and complex seabed textures like sand-ripples and sea-grass often lead to false segmentation. To tackle these problems, we propose a new method to incorporate the possible spatial correlation between neighboring pixels in the sonar image for improved segmentation. Our method modifies the expectation-maximization (EM) algorithm by adding an intermediate step (I-step) between the expectation (E-step) and maximization (M-step). Results show that our proposed method achieves improved segmentation performance over the state-of-the-art and is also robust to different seabed texture and for intensity inhomogeneity.",

keywords = "Expectation-maximization (EM), gamma distribution, object detection, sand ripples, sonar image segmentation, speckle noise",

author = "Avi Abu and Roee Diamant",

note = "Funding Information: Our second recommendation relates to financial support of students/trainees, from the undergraduate to postdoctoral levels. Programs should set aside training and travel monies for ethnic minority students/trainees. Such funds could cover tuition, stipends, and professional development to allow for students/trainees to attend national conferences, since these students may derive particular benefit from connecting with ethnic minority students in other programs and from connecting with a nationwide network of ethnic minority neuropsychologists present at these meetings. Publisher Copyright: {\textcopyright} 2017 IEEE.; OCEANS 2017 - Aberdeen ; Conference date: 19-06-2017 Through 22-06-2017",

year = "2017",

month = oct,

day = "25",

doi = "10.1109/OCEANSE.2017.8084853",

language = "English",

series = "OCEANS 2017 - Aberdeen",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "1--5",

booktitle = "OCEANS 2017 - Aberdeen",

address = "United States",

}

TY - GEN

T1 - Unsupervised segmentation of underwater objects in sonar images

AU - Abu, Avi

AU - Diamant, Roee

N1 - Funding Information: Our second recommendation relates to financial support of students/trainees, from the undergraduate to postdoctoral levels. Programs should set aside training and travel monies for ethnic minority students/trainees. Such funds could cover tuition, stipends, and professional development to allow for students/trainees to attend national conferences, since these students may derive particular benefit from connecting with ethnic minority students in other programs and from connecting with a nationwide network of ethnic minority neuropsychologists present at these meetings. Publisher Copyright: © 2017 IEEE.

PY - 2017/10/25

Y1 - 2017/10/25

N2 - We focus on the segmentation of sonar images for the aim of underwater object detection. Speckle noise and intensity inhomogeneity may cause false detections, and complex seabed textures like sand-ripples and sea-grass often lead to false segmentation. To tackle these problems, we propose a new method to incorporate the possible spatial correlation between neighboring pixels in the sonar image for improved segmentation. Our method modifies the expectation-maximization (EM) algorithm by adding an intermediate step (I-step) between the expectation (E-step) and maximization (M-step). Results show that our proposed method achieves improved segmentation performance over the state-of-the-art and is also robust to different seabed texture and for intensity inhomogeneity.

AB - We focus on the segmentation of sonar images for the aim of underwater object detection. Speckle noise and intensity inhomogeneity may cause false detections, and complex seabed textures like sand-ripples and sea-grass often lead to false segmentation. To tackle these problems, we propose a new method to incorporate the possible spatial correlation between neighboring pixels in the sonar image for improved segmentation. Our method modifies the expectation-maximization (EM) algorithm by adding an intermediate step (I-step) between the expectation (E-step) and maximization (M-step). Results show that our proposed method achieves improved segmentation performance over the state-of-the-art and is also robust to different seabed texture and for intensity inhomogeneity.

KW - Expectation-maximization (EM)

KW - gamma distribution

KW - object detection

KW - sand ripples

KW - sonar image segmentation

KW - speckle noise

UR - http://www.scopus.com/inward/record.url?scp=85044580943&partnerID=8YFLogxK

U2 - 10.1109/OCEANSE.2017.8084853

DO - 10.1109/OCEANSE.2017.8084853

M3 - Conference contribution

AN - SCOPUS:85044580943

T3 - OCEANS 2017 - Aberdeen

SP - 1

EP - 5

BT - OCEANS 2017 - Aberdeen

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - OCEANS 2017 - Aberdeen

Y2 - 19 June 2017 through 22 June 2017

ER -

Unsupervised segmentation of underwater objects in sonar images

Abstract

Publication series

Conference

Bibliographical note

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this