Data Hiding in Compressed Video using Motion Vector

Abstract
Video data hiding is still an important research topic due to the design complexities involved. It propose a new video data hiding method that makes use of erasure correction capability of Repeat Accumulate codes and superiority of Forbidden Zone Data Hiding. Selective embedding is utilized in the proposed method to determine host signal samples suitable for data hiding. Video data hiding method also contains a temporal synchronization scheme in order to withstand frame drop and insert attacks. The proposed framework is tested by typical broadcast material against MPEG-2, H.264 compression, frame-rate conversion attacks, as well as other well-known video data hiding methods. The decoding error values are reported for typical system parameters. The simulation results indicate that the framework can be successfully utilized in video data hiding applications.
Keywords:Data hiding, Encryption, Decryption Stegnography, Steganalysis, MotionVector.
INTRODUCTION
Data hiding is the process of embedding information into a host medium. In general, visual media are preferred due to their wide presence and the tolerance of human perceptual systems involved. Although the general structure of data hiding process does not depend on the host media type, the methods vary depending on the nature of such media. For instance, image and video data hiding share many common points; however video data hiding necessitates more complex designs as a result of the additional temporal dimension. Therefore, video data hiding continues to constitute an active research area. Data hiding in video sequences is performed in two major ways: bit stream-level and data-level. In bit stream-level, the redundancies within the current compression standards are exploited. Typically, encoders have various options during encoding and this freedom of selection is suitable for manipulation with the aim of data hiding. However, these methods highly rely on the structure of the bit stream; hence, they are quite fragile, in the sense that in many cases they cannot survive any format conversion or transcoding even without any significant loss of perceptual quality. As a result, this type of data hiding methods is generally proposed for fragile applications, such as authentication. On the other hand, data-level methods are more robust to attacks. Therefore, they are suitable for a broader range of applications. Despite their fragility, the bit stream-based methods are still attractive for data hiding applications. For instance, in the redundancy in block size selection of H.264 encoding is exploited for hiding data. In another approach the quantization parameter and DCT (Discrete Cosine Transform) coefficients are altered in the bit stream-level. However, most of the video data hiding methods utilize uncompressed video data. Sarkar et. al proposes a high volume transform domain data hiding in MPEG-2 videos. They apply QIM to low-frequency DCT coefficients and adapt the quantization parameter based on MPEG-2 parameters. Furthermore, they vary the embedding rate depending on the type of the frame. As a result, insertions and erasures occur at the decoder, which causes resynchronization. They utilize Repeat Accumulate (RA) codes in order to withstand erasures. Since they adapt the parameters according to type of frame, each frame is processed separately RA codes are already applied in image data hiding. In adaptive block selection results in de-synchronization and they utilize RA codes to handle erasures. Insertions and erasures can be also handled by convolution codes as in. The authors use convolution codes at embedded however; the burden is placed on the decoder. Multiple parallel Viterbidecoders are used to correct resynchronization errors. However, it is observed. that such a scheme is successful when the number of selected host signal samples is much less than the total number of host signal samples. In 3-D DWT domain is used to hide data. They use LL sub band coefficients and do not perform any adaptive selection. Therefore, they do not use error correction codes robust to erasures. Instead, they use BCH code to increase error correction capability. By means of simple rules applied to the frame markers, it introduces certain level of robustness against frame drop, repeat and inserts attacks. And it also utilizes systematic RA codes to encode message bits and frame marker bits. Each bit is associated with a block residing in a group of frames. Random interleaving is performed spatio-temporally; hence, dependency to local characteristics is reduced. Host signal coefficients used for data hiding are selected at four stages. First, frame selection is performed. Frames with sufficient number of blocks are selected. Next, only some predetermined low frequency DCT coefficients are permitted to hide data. Then the average energy of the block is expected to be greater than a predetermined threshold. In the final stage, the energy of each coefficient is compared against another threshold.