A genetic algorithm for the segmentation of known touching objects

Research output: Contribution to journalArticle

7 Citations (Scopus)

Abstract

Problem statement: Segmentation is the first and fundamental step in the process of computer vision and object classification. However, complicate or similar colour pattern add complexity to the segmentation of touching objects. The objective of this study was to develop a robust technique for the automatic segmentation and classification of touching plastic bottles, whose features were previously stored in a database. Approach: Our technique was based on the possibility to separate the two objects by means of a segment of straight line, whose position was determined by a genetic approach. The initial population of the genetic algorithm was heuristically determined among a large set of cutting lines, while further generations were selected based on the likelihood of the two objects with the images stored in the database. Results: Extensive testing, which was performed on random couples out of a population of 50 bottles, showed that the correct segmentation could be achieved in success rates above 90% with only a limited number of both chromosomes and iterations, thus reducing the computing time. Conclusion: These findings proved the effectiveness of our method as far as touching plastic bottles are concerned. This technique, being absolutely general, can be extended to any situation in which the properties of single objects were previously stored in a database.

Original languageEnglish
Pages (from-to)711-716
Number of pages6
JournalJournal of Computer Science
Volume5
Issue number10
DOIs
Publication statusPublished - 2009

Fingerprint

Plastic bottles
Genetic algorithms
Bottles
Chromosomes
Computer vision
Color
Testing

Keywords

  • Computer vision
  • Genetic algorithm
  • Segmentation

ASJC Scopus subject areas

  • Software
  • Computer Networks and Communications
  • Artificial Intelligence

Cite this

@article{3d1a656cf5af42a386383751d09e3e6f,
title = "A genetic algorithm for the segmentation of known touching objects",
abstract = "Problem statement: Segmentation is the first and fundamental step in the process of computer vision and object classification. However, complicate or similar colour pattern add complexity to the segmentation of touching objects. The objective of this study was to develop a robust technique for the automatic segmentation and classification of touching plastic bottles, whose features were previously stored in a database. Approach: Our technique was based on the possibility to separate the two objects by means of a segment of straight line, whose position was determined by a genetic approach. The initial population of the genetic algorithm was heuristically determined among a large set of cutting lines, while further generations were selected based on the likelihood of the two objects with the images stored in the database. Results: Extensive testing, which was performed on random couples out of a population of 50 bottles, showed that the correct segmentation could be achieved in success rates above 90{\%} with only a limited number of both chromosomes and iterations, thus reducing the computing time. Conclusion: These findings proved the effectiveness of our method as far as touching plastic bottles are concerned. This technique, being absolutely general, can be extended to any situation in which the properties of single objects were previously stored in a database.",
keywords = "Computer vision, Genetic algorithm, Segmentation",
author = "Edgar Scavino and {Abd. Wahab}, Dzuraidah and Hassan Basri and Mustafa, {Mohd. Marzuki} and Aini Hussain",
year = "2009",
doi = "10.3844/jcssp.2009.711.716",
language = "English",
volume = "5",
pages = "711--716",
journal = "Journal of Computer Science",
issn = "1549-3636",
publisher = "Science Publications",
number = "10",

}

TY - JOUR

T1 - A genetic algorithm for the segmentation of known touching objects

AU - Scavino, Edgar

AU - Abd. Wahab, Dzuraidah

AU - Basri, Hassan

AU - Mustafa, Mohd. Marzuki

AU - Hussain, Aini

PY - 2009

Y1 - 2009

N2 - Problem statement: Segmentation is the first and fundamental step in the process of computer vision and object classification. However, complicate or similar colour pattern add complexity to the segmentation of touching objects. The objective of this study was to develop a robust technique for the automatic segmentation and classification of touching plastic bottles, whose features were previously stored in a database. Approach: Our technique was based on the possibility to separate the two objects by means of a segment of straight line, whose position was determined by a genetic approach. The initial population of the genetic algorithm was heuristically determined among a large set of cutting lines, while further generations were selected based on the likelihood of the two objects with the images stored in the database. Results: Extensive testing, which was performed on random couples out of a population of 50 bottles, showed that the correct segmentation could be achieved in success rates above 90% with only a limited number of both chromosomes and iterations, thus reducing the computing time. Conclusion: These findings proved the effectiveness of our method as far as touching plastic bottles are concerned. This technique, being absolutely general, can be extended to any situation in which the properties of single objects were previously stored in a database.

AB - Problem statement: Segmentation is the first and fundamental step in the process of computer vision and object classification. However, complicate or similar colour pattern add complexity to the segmentation of touching objects. The objective of this study was to develop a robust technique for the automatic segmentation and classification of touching plastic bottles, whose features were previously stored in a database. Approach: Our technique was based on the possibility to separate the two objects by means of a segment of straight line, whose position was determined by a genetic approach. The initial population of the genetic algorithm was heuristically determined among a large set of cutting lines, while further generations were selected based on the likelihood of the two objects with the images stored in the database. Results: Extensive testing, which was performed on random couples out of a population of 50 bottles, showed that the correct segmentation could be achieved in success rates above 90% with only a limited number of both chromosomes and iterations, thus reducing the computing time. Conclusion: These findings proved the effectiveness of our method as far as touching plastic bottles are concerned. This technique, being absolutely general, can be extended to any situation in which the properties of single objects were previously stored in a database.

KW - Computer vision

KW - Genetic algorithm

KW - Segmentation

UR - http://www.scopus.com/inward/record.url?scp=70349275731&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=70349275731&partnerID=8YFLogxK

U2 - 10.3844/jcssp.2009.711.716

DO - 10.3844/jcssp.2009.711.716

M3 - Article

AN - SCOPUS:70349275731

VL - 5

SP - 711

EP - 716

JO - Journal of Computer Science

JF - Journal of Computer Science

SN - 1549-3636

IS - 10

ER -