Discrete Cosine Transform-Domain Based Video Steganography using Multiple Object Tracking

Abstract

In the current era, the need for transmission of confidential data is pivotal in all domains of technology. New methodologies are being proposed for this very purpose, each being developed on specific shortcomings of previously proposed methods. In this paper, one such method for embedding and communicating confidential information has been proposed which employs encryption and error correction algorithms to further improve the reliability and robustness of the system. First, the information is preprocessed by encrypting the text using RC4 algorithm. The encrypted text is processed using hamming codes (4,7) to improve its reliability. After the pre-processing stage, the video is processed to identify reliable motion regions using the MOT (Multiple Object Tracking) algorithm. The identified regions are used to embed the confidential text by calculating its DCT coefficients and manipulating its values to embed the information. The non-motion region is used to store the key used during the pre-processing stage. This completes the processes at the transmitter’s part. The inverse of the process-extraction and decoding is applied at the receiver’s side to retrieve the confidential piece of information being transferred. The proposed methodology further focusses on comparing the visual quality and hiding capacity achieved by adopting DCT and DWT algorithms, to determine the better of the two frequency domain techniques in steganographic applications.