Please use this identifier to cite or link to this item: http://hdl.handle.net/10603/13651
Title: Some studies on statistical texture analysis methods and their application
Researcher: Sreeja Mole S S
Guide(s): Ganesan, L.
Keywords: Statistical texture analysis, mammogram images, local binary pattern operator, texture spectrum operator, entropy based local descriptor
Upload Date: 5-Dec-2013
University: Anna University
Completed Date: 
Abstract: The main aim of texture analysis is to capture the tonal information of the image, since texture is a major property of an image that represents important information about the arrangement of features. Texture determines the overall visual smoothness of the image. Texture is measured statistically using a sliding window through out the image. In this study, the different methods of statistical texture analysis and comparisons have been done. As an application of texture analysis in medical image classification, mammogram images have been studied. In the first study, a new statistical method of texture analysis has been presented, which is focused on texture characterization and discrimination with features like Local Binary Pattern Operator, Texture Spectrum Operator and Entropy Based Local Descriptor. The second study of this thesis presents a new classification technique named Unsupervised Hybrid Classification for Texture Analysis (UHCTA) that integrates different unsupervised methods. The third study of this thesis presents an image classification method based on Pixel by Pixel with maximum likelihood estimates that must be compared to a single window classification not only to monochrome images but with the color images too. Texture analysis has been very much used in medical image problems. The fourth study of this thesis presents the application of both supervised and unsupervised statistical method for the detection of Microcalcification on Mammogram. On the whole, the texture analysis methods proposed and used in this dissertation are very useful in providing higher classification accuracy. The results suggest that the proposed classification methods for texture analysis perform significantly better in the way of incorporating improved accuracy measures. The approach would definitely increase its usability and adds more to the future prospects of improving classification in texture analysis for many applications. newline newline newline
Pagination: xx, 143
URI: http://hdl.handle.net/10603/13651
Appears in Departments:Faculty of Information and Communication Engineering

Files in This Item:
File Description SizeFormat 
01_title.pdfAttached File49.58 kBAdobe PDFView/Open
02_certificates.pdf424.52 kBAdobe PDFView/Open
03_abstract.pdf14.73 kBAdobe PDFView/Open
04_acknowledgement.pdf13.73 kBAdobe PDFView/Open
05_contents.pdf45.35 kBAdobe PDFView/Open
06_chapter 1.pdf41.79 kBAdobe PDFView/Open
07_chapter 2.pdf76.97 kBAdobe PDFView/Open
08_chapter 3.pdf654.51 kBAdobe PDFView/Open
09_chapter 4.pdf131.83 kBAdobe PDFView/Open
10_chapter 5.pdf119.37 kBAdobe PDFView/Open
11_chapter 6.pdf641.65 kBAdobe PDFView/Open
12_chapter 7.pdf26.58 kBAdobe PDFView/Open
13_appendix 1.pdf22.69 kBAdobe PDFView/Open
14_references.pdf31.72 kBAdobe PDFView/Open
15_publications.pdf16.67 kBAdobe PDFView/Open
16_vitae.pdf12.67 kBAdobe PDFView/Open


Items in Shodhganga are protected by copyright, with all rights reserved, unless otherwise indicated.