Detecting hidden messages using higher-order statistical models

Abstract

Techniques for information hiding have become increasingly more sophisticated and widespread. With high-resolution digital images as carriers, detecting hidden messages has become considerably more difficult. This paper describes a new approach to detecting hidden messages in images. The approach uses a wavelet-like decomposition to build higher-order statistical models of natural images. A Fisher (1936) linear discriminant analysis is then used to discriminate between untouched and adulterated images.