Please use this identifier to cite or link to this item:
http://hdl.handle.net/10603/426313
Title: | Reversible data hiding in grayscale images using prediction error expansion based techniques |
Researcher: | Ravi, Uyyala. |
Guide(s): | Rajarshi Pal. |
Keywords: | Computer Science Computer Science Software Engineering Engineering and Technology |
University: | University of Hyderabad |
Completed Date: | 2020 |
Abstract: | Reversible data hiding is a special kind of data hiding technique, where newlineoriginal cover media can be restored along with extraction of hidden newlinedata. In this thesis, reversible data hiding is discussed in the context newlineof grayscale image as cover media. Several types of reversible data newlinehiding techniques exist in literature. Prediction error expansion based newlinereversible data hiding techniques exhibit superiority in performance newlineover other types of reversible data hiding techniques. In a prediction newlineerror expansion based technique, a pixel value is predicted using a newlinepixel prediction strategy. Then, data bit is added in the expanded newlineprediction error of a pixel. A good pixel prediction strategy is key newlineto this technique. A small prediction error leads to less embedding newlinedistortion. newlineIn this thesis, several novel reversible data hiding techniques are proposed newlineby exploiting several strategies for pixel value prediction. In newlinethe first of the proposed techniques, B-tree triangular decomposition newlinetechnique is used to obtain a set of reference pixels. Non-reference newlinepixel values are interpolated (predicted) using these reference pixel newlinevalues. In the second of the proposed techniques, reference pixels are newlinerandomly distributed throughout an image. Non-reference pixel values newlineare predicted using weighted median of the values at the nearby newlinereference pixels. In the third of the proposed technique, a pixel value is newlinepredicted as an average of few linearly predicted values in the selected newlinedirectional contexts. A few directions are selected by analyzing the newlinepixel values in an 8-neighborhood of the pixel. Similarly, in the fourth newlineof the proposed techniques, gradient estimations at several directions newlineare used to select directional contexts. Then, a weighted average of newlinevi newlinetwo linearly predicted values in the selected directions provides the newlinefinal predicted value of the pixel. Finally, the performances of several newlineneighborhood-based and gradient-based predictors are compared newlineto highlight the need of a multi-predictor strategy. A novel multipredictor newline |
Pagination: | 160p |
URI: | http://hdl.handle.net/10603/426313 |
Appears in Departments: | Department of Computer & Information Sciences |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
80_recommendation.pdf | Attached File | 13.58 MB | Adobe PDF | View/Open |
abstract.pdf | 73.36 kB | Adobe PDF | View/Open | |
annexures.pdf | 288.61 kB | Adobe PDF | View/Open | |
chapter 1.pdf | 363.92 kB | Adobe PDF | View/Open | |
chapter 2.pdf | 278.81 kB | Adobe PDF | View/Open | |
chapter 3.pdf | 1.28 MB | Adobe PDF | View/Open | |
chapter 4.pdf | 753.84 kB | Adobe PDF | View/Open | |
chapter 5.pdf | 1.06 MB | Adobe PDF | View/Open | |
chapter 6.pdf | 728.07 kB | Adobe PDF | View/Open | |
chapter 7.pdf | 10.02 MB | Adobe PDF | View/Open | |
chapter 8.pdf | 106.07 kB | Adobe PDF | View/Open | |
contents.pdf | 84.76 kB | Adobe PDF | View/Open | |
prelim pages.pdf | 264.73 kB | Adobe PDF | View/Open | |
title.pdf | 309.01 kB | Adobe PDF | View/Open |
Items in Shodhganga are licensed under Creative Commons Licence Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0).
Altmetric Badge: