A Cumulative Sums Algorithm for Segmentation of Digital X-ray Images

S. E. Vorobeychikov, S. V. Chakhlov, V. A. Udod

Research output: Contribution to journalArticle

Abstract

A mathematical model of digital X-ray image of the test object is presented for the case when the main type of image distortion is the noise due to the quantum nature of the radiation. The new multilevel cumulative sums algorithm for automatic image segmentation is proposed. The algorithm is based on the edge detection of the segments homogeneous in brightness along the image rows and columns by repeatedly applying the cumulative sums procedure. The efficiency of the proposed algorithm is compared with the known threshold and Leader algorithms. The comparison was performed on simulated images as well as on the X-ray image of a weld. The mean square errors for the new algorithm were about two and three times less than for the threshold and Leader algorithms, correspondingly.

Original languageEnglish
Article number78
JournalJournal of Nondestructive Evaluation
Volume38
Issue number3
DOIs
Publication statusPublished - 1 Sep 2019

Fingerprint

X rays
Edge detection
Image segmentation
Mean square error
Luminance
Welds
Mathematical models
Radiation

Keywords

  • Digital X-ray image
  • Inspected object
  • Mathematical model
  • Segmentation algorithms

ASJC Scopus subject areas

  • Mechanics of Materials
  • Mechanical Engineering

Cite this

A Cumulative Sums Algorithm for Segmentation of Digital X-ray Images. / Vorobeychikov, S. E.; Chakhlov, S. V.; Udod, V. A.

In: Journal of Nondestructive Evaluation, Vol. 38, No. 3, 78, 01.09.2019.

Research output: Contribution to journalArticle

@article{4cbaba7cdbed4fdcaa95e91a261944d2,
title = "A Cumulative Sums Algorithm for Segmentation of Digital X-ray Images",
abstract = "A mathematical model of digital X-ray image of the test object is presented for the case when the main type of image distortion is the noise due to the quantum nature of the radiation. The new multilevel cumulative sums algorithm for automatic image segmentation is proposed. The algorithm is based on the edge detection of the segments homogeneous in brightness along the image rows and columns by repeatedly applying the cumulative sums procedure. The efficiency of the proposed algorithm is compared with the known threshold and Leader algorithms. The comparison was performed on simulated images as well as on the X-ray image of a weld. The mean square errors for the new algorithm were about two and three times less than for the threshold and Leader algorithms, correspondingly.",
keywords = "Digital X-ray image, Inspected object, Mathematical model, Segmentation algorithms",
author = "Vorobeychikov, {S. E.} and Chakhlov, {S. V.} and Udod, {V. A.}",
year = "2019",
month = "9",
day = "1",
doi = "10.1007/s10921-019-0616-3",
language = "English",
volume = "38",
journal = "Journal of Nondestructive Evaluation",
issn = "0195-9298",
publisher = "Springer New York",
number = "3",

}

TY - JOUR

T1 - A Cumulative Sums Algorithm for Segmentation of Digital X-ray Images

AU - Vorobeychikov, S. E.

AU - Chakhlov, S. V.

AU - Udod, V. A.

PY - 2019/9/1

Y1 - 2019/9/1

N2 - A mathematical model of digital X-ray image of the test object is presented for the case when the main type of image distortion is the noise due to the quantum nature of the radiation. The new multilevel cumulative sums algorithm for automatic image segmentation is proposed. The algorithm is based on the edge detection of the segments homogeneous in brightness along the image rows and columns by repeatedly applying the cumulative sums procedure. The efficiency of the proposed algorithm is compared with the known threshold and Leader algorithms. The comparison was performed on simulated images as well as on the X-ray image of a weld. The mean square errors for the new algorithm were about two and three times less than for the threshold and Leader algorithms, correspondingly.

AB - A mathematical model of digital X-ray image of the test object is presented for the case when the main type of image distortion is the noise due to the quantum nature of the radiation. The new multilevel cumulative sums algorithm for automatic image segmentation is proposed. The algorithm is based on the edge detection of the segments homogeneous in brightness along the image rows and columns by repeatedly applying the cumulative sums procedure. The efficiency of the proposed algorithm is compared with the known threshold and Leader algorithms. The comparison was performed on simulated images as well as on the X-ray image of a weld. The mean square errors for the new algorithm were about two and three times less than for the threshold and Leader algorithms, correspondingly.

KW - Digital X-ray image

KW - Inspected object

KW - Mathematical model

KW - Segmentation algorithms

UR - http://www.scopus.com/inward/record.url?scp=85070936222&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85070936222&partnerID=8YFLogxK

U2 - 10.1007/s10921-019-0616-3

DO - 10.1007/s10921-019-0616-3

M3 - Article

VL - 38

JO - Journal of Nondestructive Evaluation

JF - Journal of Nondestructive Evaluation

SN - 0195-9298

IS - 3

M1 - 78

ER -