Masters Theses

Abstract

"Target segmentation and detection problem is one of the preliminary approaches in analyzing an image. Many edge, region and texture based techniques have been proposed in literature. These techniques often have limitations based on types of image data, kind of segmentation of interest and the problem domain. Mine detection problem dealt with for this thesis has mine target images which have a center target consisting of false alarm or mine patch which has to be detected and features evaluated to evaluate various parameter performances. Thus, segmenting the mine patch becomes the first step of analyzing these images. This thesis is divided into two parts. First part deals with segmenting the single and multi-band mine linages for target detection. An immersion-based watershed approach is adopted with median filtering to obtain a continuous image and thus a better segmentation. The over segmentation caused by watershed algorithm is addressed by using the region dissimilarity criterion for region merging. The second part of tills thesis consists of discussion about various spectral features devised to identify the "mineness" of the target detected thus aiding in determining if the target is a mine or a false alarm. False alarm study is one of the major parts in mine detection as there are many false alarm signatures similar to mines. This is discussed in more detail and segmentation results and Receiver Operating Curves (ROCs) for false alarm study is presented wherever relevant"--Abstract, page iii.

Advisor(s)

Agarwal, Sanjeev, 1971-

Committee Member(s)

Moss, Randy Hays, 1953-
Shrestha, Bijaya

Department(s)

Electrical and Computer Engineering

Degree Name

M.S. in Computer Engineering

Sponsor(s)

Night Vision Laboratory. Countermine Division

Publisher

Missouri University of Science and Technology

Publication Date

2010

Pagination

ix, 46 pages

Note about bibliography

Includes bibliographical references (pages 41-45).

Rights

© 2010 Deepika Kamakshi Timmavajjula, All rights reserved.

Document Type

Thesis - Open Access

File Type

text

Language

English

Subject Headings

Image processing -- Computer programsMines (Military explosives) -- DetectionReceiver operating characteristic curves -- Computer programsImage processing -- Mathematics

Thesis Number

T 10263

Print OCLC #

870997789

Electronic OCLC #

909616820

Download

Share

 
COinS