Login | Register

GPIS: genetic programming based image segmentation with applications to biomedical object detection

Title:

GPIS: genetic programming based image segmentation with applications to biomedical object detection

Dhot, Tarundeep Singh (2009) GPIS: genetic programming based image segmentation with applications to biomedical object detection. Masters thesis, Concordia University.

[img]
Preview
Text (application/pdf)
MR63236.pdf - Accepted Version
4MB

Abstract

Image segmentation plays a critical role in many image analysis applications. However, it is ill-defined in nature and remains one of the most intractable problems in image processing. In this thesis, we propose a genetic programming based algorithm for image segmentation (GPIS). Typically, genetic programming is a Darwinian-evolution inspired program discovery method and in the past it has been successfully used as an automatic programming tool. We make use of this property of GP to evolve efficient and accurate image segmentation programs from a pool of basic image analysis operators. In addition, we provide no a priori information about that nature of the images to the GP. The algorithm was tested on two separate medical image databases and results show the proposed GP's ability to adapt and produce short and accurate segmentation algorithms, irrespective of the database in use. We compared our results with a popular GA based image segmentation/classification system, GENIE Pro. We found that our proposed algorithm produced accurate image segmentations performed consistently on both databases and could possibly be extended to other image databases as a general-purpose image segmentation tool.

Divisions:Concordia University > Gina Cody School of Engineering and Computer Science > Electrical and Computer Engineering
Item Type:Thesis (Masters)
Authors:Dhot, Tarundeep Singh
Pagination:xv, 106 leaves : ill. ; 29 cm.
Institution:Concordia University
Degree Name:M.A. Sc.
Program:Electrical and Computer Engineering
Date:2009
Thesis Supervisor(s):Kharma, N
ID Code:976222
Deposited By: Concordia University Library
Deposited On:22 Jan 2013 16:21
Last Modified:18 Jan 2018 17:42
Related URLs:
All items in Spectrum are protected by copyright, with all rights reserved. The use of items is governed by Spectrum's terms of access.

Repository Staff Only: item control page

Downloads per month over past year

Back to top Back to top