Models of heterogeneous covers for quantitative steganalysis

Abstract

Unlike the other chapters in this part of the book, this one does not present a new or improved detector, but contributes a methodology to deal with sources of heterogeneous covers in the context of steganalysis. In general, the detectability of secret messages, and hence the success of steganalysis, depends on many factors, which can be attributed to 1. the embedding function Embed and (implicitly) the embedding domain, or 2. the detection method Detect, or 3. the message m (for uniform random bit strings, mainly its length matters), or finally 4. the cover x(0) (e.g., properties related to source, size, preprocessing artefacts, etc.).