Data Hiding based on Independent Component Analyses (ICA)

Abstract

Nowadays, cases of theft of important data both by employees of the organization and outside hackers are increasing day-by-day. So, new methods for information hiding and secret communication are need of today. Steganography is an option for it. Embedding a secret message into other meaningful messages (cover media) without disturbing the features of the cover media is known as steganography [1]. The first method proposed is called the cocktail mixed method. In this method hiding a secret image inside the audio cover. Using Discrete Wavelet Transform (DWT) and Discrete cosine transform (DCT) for the cover file while secret information is an image encrypted with a Henon map method to increase the complexity. The resulting stego signal mixed with other audio to increase complexity. The second method proposed is called the Single Value Decomposition (SVD) method. This method used mixing two matrices of the three matrices results from applying SVD on input binary image and embedding these parameters in the audio signal and the third matrix use as a key for extraction.  The DWT is applied to the audio file for separating low frequency from the high-frequency signal. The third method proposed is called a speech hiding method. That hides speech signals in audio signals after compressing the speech using a discrete cosine transform (DCT) and another speech is used as a secret key and applying the same operations on it. The cover audio signal is segmented into frames and transform into the frequency domain using Fast Fourier Transform (FFT). Finally, the results proved the quality of the proposed methods and their strength against attacks. The mean secured error, signal to noise ratio, and accuracy for retrieval images after embedding it gave good results.