Automated Detection of Semagram-Laden Images Using Adaptive Neural Networks
Proceedings of SPIE
ISSN or ISBN
Digital steganography has been used extensively for electronic copyright stamping, but also for criminal or covert activities. While a variety of techniques exist for detecting steganography the identification of semagrams, messages transmitted visually in a non-textual format remain elusive. The work that will be presented describes the creation of a novel application which uses hierarchical neural network architectures to detect the likely presence of a semagram message in an image. The application was used to detect semagrams containing Morse Code messages with over 80% accuracy. These preliminary results indicate a significant advance in the detection of complex semagram patterns.
Cannady, James D. Jr. and Cerkez, Paul S., "Automated Detection of Semagram-Laden Images Using Adaptive Neural Networks" (2010). CEC Faculty Articles. 446.