A Steganographic Framework for Reference Colour Based Encoding and Cover Image Selection

This is a small version of a paper published at the ISW2000 in Wollongong describing only the new codec for embedding and extraction of the additional information and some experimental results. 1 An adaptive steganographic framework In this section, we introduce a new hybrid steganographic approach. The presented framework is distinguished from previous approaches by a greater flexibility facing changes of the demands. This flexibility can be reached on two different levels: (1) On the level of the used systems and algorithms, and (2) on the level of a concrete implementation. The first point can be realised during the design phase and implementation of the steganographic system by exchanging of modules. On the level of a concrete implementation, changes are possible during the execution time of the system. The basic idea of the new approach is to use so called reference colours, which form the starting point for the embedding and extraction of the additional information. These reference colours are chosen such that if the cover image is described only by these colours, there are no visual differences to the original cover image. Starting from an image that is normalised in this way, parts of information can be embedded into it by small changes of the reference colours . Here the discussed effect of just noticeable differences is used. The extraction itself is based on the comparison between the reference colours and the colour information of the stego image. Therefore, sender and receiver must share the same reference colour information to extract the information correctly. In image domain the unequivocal assignment and decision of membership for pixel values to parts of the additional information is based on a Pseudo-Noise-Function1. The used cover, the reference colours of the cover and the parameters, which steer the random function can be suitably chosen depending on the existing requirements. The presented framework consists of modules which can be chosen according to the suitability of components for embedding and extracting (Fig. 1). According to the properties of the individual requirements (cf. section ??) a suitable candidate can be selected for each codec step. To embed a message, we first have to chose a suitable cover (module 1) which depends on the user requirements and is based on the prepositions given in section ??. After that, we transform this cover in a more suitable colour system 1 Implementation of Kerkhoff’s principle [Sim83].