A Multi-level Steganography System Using Wavelet Transform Technique

Techniques and applications for information hiding have become increasingly more sophisticated and widespread. In a digital world, steganography is introduced to hide the existence of the communication by concealing a secret message file inside another unsuspicious cover file.
The prime fundamental objective achieved in this thesis, is to implement an efficient Multi-Level Steganography (MLS) system to achieve high level of security. In this type of systems, two levels of hiding process are implemented and two levels of extracting process are required. Thus, the secret message becomes very difficult to reconstruct. This complexity in retrieving the hidden data adds additional layer of protecting and makes it more successful.
Two main approaches of steganography have been implemented and developed in this work, they are: steganography and multi-level steganography. Four systems are proposed, two systems are implemented using steganography approach and two systems are implemented using multi-level steganography approach. These proposed systems are implemented using the ransformation domain to increase the robustness of the system, based on wavelet and multiwavelet transforms. The singular value decomposition technique is used to enhance the performance of embedding scheme, and it also offers minimum or no distortion. The covering files used to implement all proposed systems, are video files. The main reasons of choosing such covering files to host the data, are due to their complexity of detecting the presence of hidden messages and due to their relative big sizes that help in hiding a large quantity of data of many secret messages types. These mentioned characteristics propose the video files to be good covers in steganography system, which can withstand many types of attacks and at the end, full filled the aim of secret communication.
The imperceptibility of the resulted stego-cover (Cover containing the secret data) and the quality of the reconstructed message are assessed by objective and subjective fidelity criteria. The objective results show that, the largest PSNR value reached to 79 dB. The simulation results under certain parameters selection, prove that, the stego-cover and the reconstructed image are showed excellent results and achieve the main steganographic goals.