Advertisement

Framework for Data Hiding Operation Using Motion Vectors for Effective Imperceptibility Performance

  • K. Manjunath KamathEmail author
  • R. Sanjeev Kunte
Conference paper
  • 213 Downloads
Part of the Lecture Notes on Data Engineering and Communications Technologies book series (LNDECT, volume 44)

Abstract

Data Hiding is one of the frequently used security approaches for safeguarding the sensitive information of the available data as well as to transmit secret information among different ends in a vulnerable network. However, majority of data hiding scheme evolved till date is focused on its embedding capacity or else focused on introducing the distinct parameters of encryption. However, all these approaches will not only make the embedded file bulky but also, they will lose its imperceptibility characteristics. Therefore, the proposed paper introduces a simple and robust reversible data hiding process where a secret image is embedded within a video as a cover image. Motion prediction and histogram shifting approach is also utilized for obtaining highly secured bit-streams. The outcome of the study shows that the proposed system offers a better signal quality and retains maximum imperceptibility irrespective of the size of the secret image.

Keywords

Reversible data hiding Image imperceptibility Video encoding Secret image Compression 

1 Introduction

With the recent progressive growth of multimedia technologies, almost all the commercial applications have started using it in a commercial way. In this regard, the usage of various multimedia file system are mainly shared and exchanged by the user from a practical viewpoint. Various studies have already been carried out to prove that the multimedia security still remains as a big challenge to overcome [1, 2, 3, 4, 5, 6]. Hence, security approaches have been evolving in order to counter-attack the threats [6, 7, 8, 9, 10]. One of the most frequently used counter-mechanisms is data hiding scheme where a secret data is embedded for security purpose. The embedding of information, in most cases, causes the carrier to lose a part of the data, so the carrier cannot be completely recovered after it gets extracted. In some special applications, such as in the fields of medicine, military, and law, false positives would be caused even by slight distortion of a digital image. Therefore, any irreversible loss of carrier data is not allowed. Basically, Reversible Data Hiding scheme is capable of extracting the actual set of information from the given source ensuring the highest quality of information i.e. zero loss of data. Therefore, this topic is currently a center of attention for various researchers. Therefore, the proposed system introduces a very simple and novel reversible data hiding process by considering video as a cover file, which mainly targets on accomplishing the imperceptibility of the embedded secret image. The organization of the paper is as follows – Sect. 2 briefs of existing studies while research problem is briefed in Sect. 3. Adopted methodology and system design is discussed in Sect. 4 and Sect. 5 respectively. Result analysis is discussed in Sect. 6 while conclusion is briefed in Sect. 7.

2 Related Work

This section discusses about recent research work towards reversible data hiding. The most recent work is carried out by Puteaux and Puech [11] where a predictive scheme of most significant bit is used emphasizing on improving the high capacity. Qian et al. [12] have used a dual embedding scheme towards generating an encrypted bitstream. Qian et al. [13] have also used an iterative process of data hiding where first the encryption of image is carried out followed by embedding on extra information for obtained ciphered image into the server. A unique three-dimensional data hiding process using mesh framework was introduced by Jiang et al. [14]. According to the scheme, the vertex coordinates are used for mapping the integers from the decimals followed by usage of the least significant bits. Usage of homomorphic encryption is carried out by Xiang and Luo [15] along with the usage of Paillier encryption approach. The work of Yi and Zhou [16] has used a labeling scheme based on binary tree structure over the image pixels in order to facilitate encryption over image. Chen et al. [17] have used sorting of pixel approach as well as extension of the errors caused due to prediction for data encryption mechanism using directional property of the predictor. The work of Wu et al. [18] has adopted a color partitioning process for developing a unique data hiding scheme. Usage of motion vector over video encoded by H264 is carried out by Niu et al. [19]. The study discussed by Hou et al. [20] have considered the distortion problem and used reversible steganography for data hiding. Xiong et al. [21] have used homomorphic encryption while Wang et al. [22] have used histogram shifting approach for data hiding. The work of Zhang [23] used have optimal value of host data where pixel and auxiliary information is used for estimating errors. Many researchers have proposed theories for video steganography as well as compressed domain reversible video steganography using various conventional approaches of compression. Steganography system proposed by Hu et al. [24] is based on non-uniform rectangular partition uses an uncompressed domain. It uses secret video to hide inside a cover video, and the size of cover video and secret video are of same size. Similarly, various others authors e.g. Ni et al. [25], Hong et al. [26], Zhang et al. [27, 30], Ma et al. [28], and Shanableh et al. [29] have also carried out work towards addressing the problem of data hiding scheme where different conventional methodologies are applied for data hiding operation. The next section outlines the research problem.

3 Problem Description

From the review of the existing system, it can be seen that there are various forms of solutions towards strengthening the process of reversible data hiding procedure. Almost all schemes that are recently evolved are very unique and distinct from each other where the performance is found to offer a better data hiding scheme. However, a thorough investigation of its performance shows that existing system is not claimed to offer a good balance between the image quality and embedding policies. Not all work has actually emphasized on imperceptibility concept that is mandatory in data hiding scheme. Apart from this, the studies towards data hiding scheme over video is quite a less to found. The few work carried out towards video encoding scheme and data hiding actually doesn’t consider imperceptibility factor associated with reversible data hiding.

4 Proposed Methodology

The main idea of the proposed system is to offer a cost effective data hiding scheme where a multimedia file system is considered as a cover file. The proposed system implements an analytical research methodology where a unique operation flow is constructed in order to obtain a secured bit-stream of data (Fig. 1). Another interesting part of the proposed methodology is that it uses the process of video encoding mechanism, which is a part of signal compression approach where a balance between compression and data security is achieved.
Fig. 1.

Operational flow of proposed system

A closer look into Fig. 1 shows that proposed system uses the concept of motion estimation followed by prediction operation where histogram shifting operation is carried out in order to obtain reference P frame. It also applies discrete cosine transform for better compression performing after sub-sampling is carried out. The complete operation leads to generation of an encoded video where data is hidden. An illustrative discussion of this method is carried out in next section.

5 System Implementation

The proposed approach deals with embedding and extraction of the secret image pixels into a MPEG2 compressed video using Histogram shifting method in motion vectors. The cover video considered in this work is an MPEG2 compressed video file. This system presents an efficient way of transfer of information from sender to the receiver as data is hidden in the motion vectors of the selected frames. An uncompressed video is selected as cover video to hide secret image. Initially the frames are extracted and Group of Pictures (GOP) is formed from the extracted frames [11]. These frames are subjected to YCbCr color conversion. The proposed system uses luminance value obtained from the red component while other color components i.e. U and V are obtained from green and blue components respectively. In the next stage reduction of the resolution is done using Chroma subsampling. The proposed system then applies the concept of motion compensation for the purpose of minimizing the redundancy factor with respect to the temporal data in it while this operation is followed by applying standard discrete cosine transform scheme for obtaining better compression outcome. Finally, quantization operation is applied on the top of this process for effective compression as well as for better control of redundancy factor with respect to spatial data. At the end, the proposed system applies a lossless compression scheme of run length encoding for achieving better encoding performance and further Huffman encoding scheme is also continued on it. At the receiver side the process has to be reversed that is called as decoding. In decoding stage the de quantization of the data is to be carried out. The proposed system uses motion vectors where the chunks of the ciphered data is basically hidden in the form of P frame and this operation is carried out only after histogram shifting is done. During the compression, the secret image bits are extracted from the corresponding motion vectors and histogram is shifted back. This achieves reversibility characteristics. This section discusses about the algorithm implemented for this purpose.

5.1 Algorithm for Reversible Data Hiding

The steps of the proposed system are as follows: The proposed algorithm takes the input of cover video (I), which is then used for framing up group of pictures Gop (Line-1). After digitizing the secret image (which is required to be hidden), the next step is to distinguish all the obtained frames into different macroblock size (Line-2). The proposed system then performs following operation on all the macro-blocks (Line-3): a standard discrete cosine transform (DCT) is applied on the macro-blocks from the spatial to frequency domain (Line-4) which is followed by applying standard quantization technique over the obtained matrix of DCT (Line-5). This process leads to generation of a dedicate motion vector obtained for all the values of P-frames (Line-7). The next process is to apply Histogram Shifting (HS) method for the generated motion vector for each P-frame (Line-8). A conditional statement is constructed (Line-9) which checks of the bits to be embedded is equivalent to 1. In case of positive scenario, the value of the pixel is maximized to 1 (Line-10) otherwise it is retained as it is (Line-12). This operation is iterated until a stopping criteria is met which is to check if the bits of complete secret image is actually populated in the destined motion vector corresponding to all the P-frame. The proposed system than apply Entropy encoding so that data is read in zig-zag order (Line-13) followed applying Huffman encoding to reduce size. Finally, the system stores the bit stream in a file.

The proposed system uses MPEG2 standard that supports motion vectors of different block sizes compared to AVI files. Each macroblock can be divided into different modes. Owing to availability of enriched and varied number of motion vectors in MPEG2, it is seen that videos which have rich texture contain more blocks and more motion vectors. The process for extracting the hidden image from the encoded video is just reverse of the algorithmic steps of encoding (Fig. 2). In order to initiate decoding, the proposed system. The decoding process starts with the output of the encoding that is bit stream is taken. For the embedded video bit steam the reverse Huffman and Reverse of HS process is applied. During this process secret data bits are extracted from each frame. All the obtained motion vector is restored in its legacy form followed by applying inverse operation of quantization. Finally original video is converted from YUV to RGB sequence and original bits are restored. Therefore, a very simplified process of image hiding approach over a cover video file is discussed in the present paper. The implementation idea is not only simple but also robust as it offers better and cost effective form of reversible data hiding scheme. A closer look into the algorithm implementation will also highlight that proposed system there is a significant contribution of histogram-shifting process over the presented video decoding system. The data hiding capacity is enhanced by incorporating motion vector for embedding data in the video stream of MPEG2. The decoding process is quite simple as it offers better form of reconstructed data in the lossless format. The embedded video can be decoded, and it can be reconstructed in lossless format by shifting histogram to its normal position after extraction of the secret information. Therefore, the proposed algorithm offers a cost effective mechanism to perform an effective data hiding process over video stream without using any complex encoding scheme. Figure 2 highlights the operational steps involved in decoding process.
Fig. 2.

Block diagram for the process of video decoding

6 Result Analysis

Before starting the discussion about the result obtained by the proposed system, it is essential to understand that proposed system implements MPEG2 compression approach. The MPEG2 compressed bit stream has more motion vectors than uncompressed video standards, so it is suitable to use motion vectors as embedding cover of reversible data hiding process. The proposed system presents an RDH scheme for HS to perform a wide range of applications. A large number of RDH schemes based on HS are proposed. It offers the capability of extensibility between the different domain as well as between different carriers that results in better performance enhancement of encoding operation. As for the compressed video, embedded information gained by modifying the motion vector allows for a great distortion of the contents, so most of algorithms are developed on the basis of the alteration carried out by the coefficient of DCT over histogram. In this paper, we have implemented histogram shifting process in generated motion vectors of MPEG2 video. Prior to starting actual analysis, the proposed system performs analysis of the histogram operation over a sample standard video which is subjected to standard MPEG2 encoding system. The main purpose of this process is to assess the trends of results obtained from the histogram shifting process. In the below histogram, peak value occurs for motion vector MV = 10 and zero point in the motion vector histogram occurs at motion vector MV = 6. Figure 3 shows the vector histogram of one frame of bus video clip encoded by MPEG2.
Fig. 3.

Sample histogram obtained from MPEG2 encoding

In the proposed work, there are two modules in the system called as Encoding Procedure and Extraction Procedure. In the Encoding procedure secret information bits (Image) are hidden into the Histogram Shifted portion of the motion vectors. Finally corresponding output bit stream is transmitted over the network. The embedded bits of information are obtained in the process of decoding steps with the aid of histograms respective to the motion vector, and histogram is shifted back to its original position and corresponding motion vectors are restored. So that after extraction of secret information, original cover video is restored back. This achieves reversibility. The proposed system uses standard images as the secret image which is showcased in Fig. 4. The analysis of the proposed system is also carried out over other standard video sequences e.g. city, news, carphone, mobile, foreman, container, salesman, and coastguard [31]. For each video sequence we take 170 frames. The GOP format used is (IPPPPPI). The proposed analysis selects one frame considered as a cover file where the secret image is embedded (Fig. 5). Multiple forms of other images are also considered while carrying out the analysis. The segment of histogram is used for embedding the secret data of motion vector.
Fig. 4.

(a) Lena image (b) Baboon image

Fig. 5.

(a) One frame of cover video (b) Extracted secret image

The analysis of the outcome of the embedding process is carried out with respect to Mean Squared Error (MSE) and Peak Signal-to-Noise Ratio (PSNR). The first analysis is carried out for both MSE and PSNR on secret image that is embedded within the cover video file. The analysis shows following trend – MSE increases with increase in dimensions of same secret image (Fig. 6). Nearly similar trend is also observed for PSNR (Fig. 7), where PSNR drops with increase of secret image size. Although, there is a drop of PSNR, but a closer look will show that this drop is very much negligible. Similar analysis is also carried out on other images too and nearly similar performance of MSE and PSNR is noticed with a fluctuation less than 5% in both PSNR and MSE of secret image.
Fig. 6.

Analysis of MSE of secret image

Fig. 7.

Analysis of PSNR of secret image

Fig. 8.

Analysis of MSE of cover video image

The next part of the assessment is associated with the MSE and PSNR trends on cover video file. This analysis is carried out considering different sample video where same secret image is embedded. All the three sample video considers are of different size and hence it is required to assess how the size of the secret image affects the PSNR (Fig. 9) and MSE (Fig. 8).
Fig. 9.

Analysis of PSNR of cover video image

The outcome shows that MSE is slightly degraded with different video samples, however, the fluctuation is highly less (Fig. 8). On the other hand, There is no significant degradation in PSNR value even in different video samples to prove that size of secret image doesn’t have any potential degradation. Similar assessment is carried out on different combination of sizes of secret images on same as well as on different video to find that PSNR stays in the range of 30–40 dB. Hence, the proposed study can be said to offer a better form of data hiding performance.

7 Conclusion

This research paper has presented a unique data hiding scheme which remains reversible in nature. The process lets the secret image to be efficiently hidden within a multimedia file like video in the form of motion vector. To achieve the reversibility of cover video, Histogram Shifting of motion vectors has been done before the data embedding process. The proposed technique is less computationally complex and can be adjusted according to the varying needs. The analysis has been proved that the proposed system offers good and stabilized MSE and PSNR performance. The inference of the outcome shows that the proposed system offers a good imperceptibility towards the embedded multimedia file where PSNR performance is found good. At present, the method is only applicable to the MPEG2 video sequences. In future, the work can be deployed to other video formats in the compressed video domains.

References

  1. 1.
    Mallikarjunan, K.N., Muthupriya, K., Shalinie, S.M.: A survey of distributed denial of service attack. In: 2016 10th International Conference on Intelligent Systems and Control (ISCO), Coimbatore, pp. 1–6 (2016)Google Scholar
  2. 2.
    Basha, S.H., Jaison, B.: Survey on patended and unpatented video watermarking techniques. In: 2017 4th International Conference on Advanced Computing and Communication Systems (ICACCS), Coimbatore, pp. 1–8 (2017)Google Scholar
  3. 3.
    Minemura, K., Wong, K., Phan, R.C., Tanaka, K.: A novel sketch attack for H.264/AVC format-compliant encrypted video. In: IEEE Transactions on Circuits and Systems for Video Technology, vol. 27, no. 11, pp. 2309–2321 (Nov 2017)Google Scholar
  4. 4.
    Panchal, U.H., Srivastava, R.: A comprehensive survey on digital image watermarking techniques. In: 2015 Fifth International Conference on Communication Systems and Network Technologies, Gwalior, pp. 591–595 (2015)Google Scholar
  5. 5.
    Venugopala, P.S., Jain, S., Sarojadevi, H., Chiplunkar, N.N.: Study of possible attacks on image and video watermark. In: 2016 3rd International Conference on Computing for Sustainable Global Development (INDIACom), New Delhi, pp. 3505–3510 (2016)Google Scholar
  6. 6.
    Le, T.H.N., Nguyen, K.H., Le, H.B.: Literature survey on image watermarking tools, watermark attacks and benchmarking tools. In: 2010 Second International Conferences on Advances in Multimedia, Athens, pp. 67–73 (2010)Google Scholar
  7. 7.
    Singh, S., Singh, A.K., Ghrera, S.P.: A recent survey on data hiding techniques. In: 2017 International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud) (I-SMAC), Palladam, pp. 882–886 (2017)Google Scholar
  8. 8.
    Mancy, L., Maria Celestin Vigila, S.: A survey on protection of medical images. In: 2015 International Conference on Control, Instrumentation, Communication and Computational Technologies (ICCICCT), Kumaracoil, pp. 503–506 (2015)Google Scholar
  9. 9.
    Igorevich, R.R., Yong, H., Min, D., Choi, E.: A study on multimedia security systems in video encryption. In: INC2010: 6th International Conference on Networked Computing, Gyeongju, pp. 1–5 (2010)Google Scholar
  10. 10.
    Moravčík, M., Segeč, P., Hrabovský, J., Papán, J., Uramová, J.: Survey of real-time multimedia security mechanisms. In: 2016 International Conference on Emerging eLearning Technologies and Applications (ICETA), Vysoke Tatry, pp. 233–238 (2016)Google Scholar
  11. 11.
    Puteaux, P., Puech, W.: An efficient MSB prediction-based method for high-capacity reversible data hiding in encrypted images. IEEE Trans. Inf. Forensics Secur. 13(7), 1670–1681 (2018)CrossRefGoogle Scholar
  12. 12.
    Qian, Z., Xu, H., Luo, X., Zhang, X.: New framework of reversible data hiding in encrypted JPEG bitstreams. IEEE Trans. Circuits Syst. Video Technol. 29(2), 351–362 (2019)CrossRefGoogle Scholar
  13. 13.
    Qian, Z., Zhou, H., Zhang, X., Zhang, W.: Separable reversible data hiding in encrypted jpeg bitstreams. In: IEEE Transactions on Dependable and Secure Computing, vol. 15, no. 6, pp. 1055–1067 (1 Nov–Dec 2018)Google Scholar
  14. 14.
    Jiang, R., Zhou, H., Zhang, W., Yu, N.: Reversible data hiding in encrypted three-dimensional mesh models. IEEE Trans. Multimedia 20(1), 55–67 (2018)CrossRefGoogle Scholar
  15. 15.
    Xiang, S., Luo, X.: Reversible data hiding in homomorphic encrypted domain by mirroring ciphertext group. IEEE Trans. Circuits Syst. Video Technol. 28(11), 3099–3110 (2018)MathSciNetCrossRefGoogle Scholar
  16. 16.
    Yi, S., Zhou, Y.: Separable and reversible data hiding in encrypted images using parametric binary tree labeling. IEEE Trans. Multimedia 21(1), 51–64 (2019)CrossRefGoogle Scholar
  17. 17.
    Chen, H., Ni, J., Hong, W., Chen, T.: High-fidelity reversible data hiding using directionally enclosed prediction. IEEE Signal Process. Lett. 24(5), 574–578 (2017)CrossRefGoogle Scholar
  18. 18.
    Wu, H., Shi, Y., Wang, H., Zhou, L.: Separable reversible data hiding for encrypted palette images with color partitioning and flipping verification. IEEE Trans. Circuits Syst. Video Technol. 27(8), 1620–1631 (2017)CrossRefGoogle Scholar
  19. 19.
    Niu, K., Yang, X., Zhang, Y.: A novel video reversible data hiding algorithm using motion vector for H.264/AVC. Tsinghua Sci. Technol. 22(5), 489–498 (2017)CrossRefGoogle Scholar
  20. 20.
    Hou, D., Zhang, W., Yang, Y., Yu, N.: Reversible data hiding under inconsistent distortion metrics. IEEE Trans. Image Process. 27(10), 5087–5099 (2018)MathSciNetCrossRefGoogle Scholar
  21. 21.
    Xiong, L., Dong, D., Xia, Z., Chen, X.: High-capacity reversible data hiding for encrypted multimedia data with somewhat homomorphic encryption. IEEE Access 6, 60635–60644 (2018)CrossRefGoogle Scholar
  22. 22.
    Wang, J., Ni, J., Zhang, X., Shi, Y.: Rate and distortion optimization for reversible data hiding using multiple histogram shifting. IEEE Trans. Cybern. 47(2), 315–326 (2017)Google Scholar
  23. 23.
    Zhang, X.: Reversible Data hiding with optimal value transfer. IEEE Trans. on Multi. 15(2), 316–325 (2013)CrossRefGoogle Scholar
  24. 24.
    Hu, S.D., KinTak, U.: Faculty of information technology: a novel video steganography based on non-uniform rectangular partition. In: IEEE International Conference on Computational Science and Engineering CSE/I-SPAN/IUC, pp. 57–61 (2011)Google Scholar
  25. 25.
    Ni, Z., Shi, Y., Ansari, N., Wei, S.: Reversible data hiding. IEEE Trans. Circuits Syst. Video Technol. 16(3), 354–362 (2006)CrossRefGoogle Scholar
  26. 26.
    Hong, W., Chen, T.S., Wu, H.Y.: An improved reversible data hiding in encrypted images using side match. IEEE Signal Pro. Lett. 19(4), 199–202 (2012)CrossRefGoogle Scholar
  27. 27.
    Zhang, X.: Separable Reversible data hiding in encrypted image. IEEE Trans. Inf. Forensics Security 7(2), 826–832 (2012)CrossRefGoogle Scholar
  28. 28.
    Ma, K.D., Zhang, W.M., Zhao, X.F., Yu, N., Li, F.: Reversible data hiding in encrypted images by reserving room before encryption. IEEE Trans. Inf. Forensics Security 8(3), 553–562 (2013)CrossRefGoogle Scholar
  29. 29.
    Shanableh, T.: Data Hiding in MPEG video files using multivariate regression and flexible macro block ordering. IEEE Trans. Inf. Forensics Security 7(2), 455–464 (2012)CrossRefGoogle Scholar
  30. 30.
    Zhang, X.P.: Reversible data hiding in encrypted image. IEEE Signal Process. Lett. 18(4), 255–258 (2011)CrossRefGoogle Scholar
  31. 31.

Copyright information

© Springer Nature Switzerland AG 2020

Authors and Affiliations

  1. 1.Yenepoya Institute of TechnologyMoodbidriIndia
  2. 2.JNN College of EngineeringShivamoggaIndia

Personalised recommendations