Accession Number:

ADA622733

Title:

Steganography Detection Using Entropy Measures

Descriptive Note:

Technical rept.

Corporate Author:

POLYTECHNIC UNIV OF PUERTO RICO SAN JUAN

Personal Author(s):

Report Date:

2012-11-16

Pagination or Media Count:

51.0

Abstract:

Steganography is the science of hiding the fact that some communication is taking place. In general encryption, encoding and decoding are not required to accomplish steganography. However, encryption serves as a layer of protection when steganography fails. The first objective of steganography is hiding the existence of data exchange between two parties. In order to achieve this beyond any mere manipulation of the mean or carrier, the existence of information exchange must be kept away from the reach of human radar sensors. Second, steganography must cause little or no impact on the carriers structure. The latter guarantees to prevent suspicious of some sort of manipulation. Third, carriers must outweigh message existence, i.e., it must be robust. The medium should withstand a certain level of modification before data existence is detected. Finally, the capacity of the medium should allows to handle a certain level of information before hidden information is detected. Steganography is studied taking under consideration the capabilities of detection of the information transferred. This include steganalysis, the techniques and methods used to detect steganography. The importance of the performance of steganography will lead, either from simple to more complex methods of detection. The importance of embedding matches the efficiency of the detection technique. There are two problems in steganalysis 1 detecting the existence of a hidden message and 2 decoding the message. As terrorist groups have been known to use steganography in planning their attacks, this has become an important problem of national security. This technical report is only concerned first, with embedding techniques and second, the problem of hidden message detection using steganalysis. The approach is to statistically analyse the least significant bits of each color dimension of each pixel to look for some kind of a pattern. In the absence of a hidden message this should look like random noise.

Subject Categories:

  • Information Science
  • Cybernetics

Distribution Statement:

APPROVED FOR PUBLIC RELEASE