期刊名称：International Journal of Network Security & Its Applications
印刷版ISSN：0975-2307
电子版ISSN：0974-9330
出版年度：2010
卷号：2
期号：4
DOI：10.5121/ijnsa.2010.2418
出版社：Academy & Industry Research Collaboration Center (AIRCC)
摘要：We propose a simple method for increasing the reliability of detection of LSB replacement steganography. RS Steganalysis Method (RSM), Sample Pair Method (SPM), and Least Square Method (LSM) are three of the most reliable steganalysis methods in the literature for LSB replacement steganography on digital images in spatial domain. However all these methods are prone to error. We propose a simple method to identify images which give very accurate results and images which give highly inaccurate results by these methods. The novelty of our method is that it does not require any knowledge about the cover images as the image classification is done based on certain statistical properties of the image, which are invariant with embedding. Thus it helps the steganalyst in attaching a level of confidence to the estimation he makes and thereby increasing the reliability of steganalysis of digital images. It also helps to choose between images for more secure LSB replacement steganography
