KL-sense secure image steganography

Research output: Contribution to journalArticle

9 Citations (Scopus)

Abstract

In this paper, we propose a computationally-efficient data hiding method which achieves Cachin's security criterion: zero Kullback-Liebler (KL) divergence. To preserve statistical properties of the cover medium, we swap pixels rather than modify them to hide information. We theoretically analyse the security of the proposed method from various perspectives. • Upper bounds of the KL divergence of second order statistics • The relationship between distortions in the DCT domain and embedding positions in the spatial domain • The upper bound on the conditional entropy in the DCT domain. We then subject our proposed stego method to several practical steganalysis algorithms. • Histogram based attacks • A higher-order statistics based universal steganalysis algorithm • A new learning based steganalysis that specifically for this hiding algorithm. Experimental results show that our data hiding method can prevent these statistical detection methods, when the embedding rate is less than or equal to 10%.

Original languageEnglish (US)
Pages (from-to)211-225
Number of pages15
JournalInternational Journal of Security and Networks
Volume6
Issue number4
DOIs
StatePublished - Jan 1 2011

Fingerprint

Steganography
Higher order statistics
Entropy
Pixels
Statistics

All Science Journal Classification (ASJC) codes

  • Safety, Risk, Reliability and Quality
  • Electrical and Electronic Engineering
  • Computer Networks and Communications

Cite this

@article{2e0942f4129e4deda3fed3c92e7f1c08,
title = "KL-sense secure image steganography",
abstract = "In this paper, we propose a computationally-efficient data hiding method which achieves Cachin's security criterion: zero Kullback-Liebler (KL) divergence. To preserve statistical properties of the cover medium, we swap pixels rather than modify them to hide information. We theoretically analyse the security of the proposed method from various perspectives. • Upper bounds of the KL divergence of second order statistics • The relationship between distortions in the DCT domain and embedding positions in the spatial domain • The upper bound on the conditional entropy in the DCT domain. We then subject our proposed stego method to several practical steganalysis algorithms. • Histogram based attacks • A higher-order statistics based universal steganalysis algorithm • A new learning based steganalysis that specifically for this hiding algorithm. Experimental results show that our data hiding method can prevent these statistical detection methods, when the embedding rate is less than or equal to 10{\%}.",
author = "Guoqi Luo and Koduvayur Subbalakshmi",
year = "2011",
month = "1",
day = "1",
doi = "https://doi.org/10.1504/IJSN.2011.045229",
language = "English (US)",
volume = "6",
pages = "211--225",
journal = "International Journal of Security and Networks",
issn = "1747-8405",
publisher = "Inderscience Enterprises Ltd",
number = "4",

}

KL-sense secure image steganography. / Luo, Guoqi; Subbalakshmi, Koduvayur.

In: International Journal of Security and Networks, Vol. 6, No. 4, 01.01.2011, p. 211-225.

Research output: Contribution to journalArticle

TY - JOUR

T1 - KL-sense secure image steganography

AU - Luo, Guoqi

AU - Subbalakshmi, Koduvayur

PY - 2011/1/1

Y1 - 2011/1/1

N2 - In this paper, we propose a computationally-efficient data hiding method which achieves Cachin's security criterion: zero Kullback-Liebler (KL) divergence. To preserve statistical properties of the cover medium, we swap pixels rather than modify them to hide information. We theoretically analyse the security of the proposed method from various perspectives. • Upper bounds of the KL divergence of second order statistics • The relationship between distortions in the DCT domain and embedding positions in the spatial domain • The upper bound on the conditional entropy in the DCT domain. We then subject our proposed stego method to several practical steganalysis algorithms. • Histogram based attacks • A higher-order statistics based universal steganalysis algorithm • A new learning based steganalysis that specifically for this hiding algorithm. Experimental results show that our data hiding method can prevent these statistical detection methods, when the embedding rate is less than or equal to 10%.

AB - In this paper, we propose a computationally-efficient data hiding method which achieves Cachin's security criterion: zero Kullback-Liebler (KL) divergence. To preserve statistical properties of the cover medium, we swap pixels rather than modify them to hide information. We theoretically analyse the security of the proposed method from various perspectives. • Upper bounds of the KL divergence of second order statistics • The relationship between distortions in the DCT domain and embedding positions in the spatial domain • The upper bound on the conditional entropy in the DCT domain. We then subject our proposed stego method to several practical steganalysis algorithms. • Histogram based attacks • A higher-order statistics based universal steganalysis algorithm • A new learning based steganalysis that specifically for this hiding algorithm. Experimental results show that our data hiding method can prevent these statistical detection methods, when the embedding rate is less than or equal to 10%.

UR - http://www.scopus.com/inward/record.url?scp=84857206460&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84857206460&partnerID=8YFLogxK

U2 - https://doi.org/10.1504/IJSN.2011.045229

DO - https://doi.org/10.1504/IJSN.2011.045229

M3 - Article

VL - 6

SP - 211

EP - 225

JO - International Journal of Security and Networks

JF - International Journal of Security and Networks

SN - 1747-8405

IS - 4

ER -