Modern telecommunication systems have seen an increased demand on delivering new types of digital media such as extended reality. The currently rolled out fifth generation mobile networks are expected to support virtual reality, augmented reality, and other immersive media. On the other hand, securing communication to prevent access from adversaries and hiding secret information have become an integral part of digital communications. In particular, steganography is a class of data hiding that provides covert communication between two parties such that the information exchange cannot be observed by an attacker. Because the human auditory system and human visual system are relatively insensitive to small changes in digital media, hiding secret information in digital media has increasingly been used. This thesis focuses on data hiding in 360^o videos which offer large resolutions that can be used for hiding secret data. To keep the computational load of data hiding low, least significant bit (LSB) data hiding methods are considered and their performance is assessed in terms of capacity and imperceptibility. The proposed LSB data hiding methods account for the human viewing behavior of watching 360^o videos on head-mounted displays (HMDs).

This thesis is divided into an introduction part and a research part based on four peer-reviewed publications. The introduction provides fundamentals of data hiding, terminologies used in data hiding, LSB data hiding concepts, and performance measures used to assess data hiding methods. The first paper in the research part provides a survey on LSB data hiding in digital audio, images, videos, and three-dimensional (3D) media. The survey shows the tremendous potential of LSB data hiding in digital media and may assist in developing novel applications based on suitable performance trade-offs between data hiding attributes. It also reveals that LSB data hiding in 3D media such as 360^o videos is not as developed as for conventional digital media. The second paper proposes an LSB data hiding method for 360^o videos which takes into account that humans pay more attention to the equator region compared to the poles when viewing 360^o videos on an HMD. The third paper presents a novel viewing direction based LSB data hiding method for 360^o videos. The distributions of viewing direction frequency for latitude and longitude are used to control the amount of secret data to be hidden at the latitude, longitude, or both latitude and longitude of 360^o  videos. Analytical expressions for the capacity offered by this method are derived and imperceptibility is assessed through fidelity and quality metrics. The fourth paper proposes a viewing direction weighted bit plane LSB data hiding method for 360^o videos that uses normalized Gaussian mixture models to control the amount of secret data and the number of bit planes used for data hiding in the latitude and longitude.

In this paper, we propose a spherical light-weight data hiding technique for 360-degree videos with equirectangular projection between sphere and plane. In particular, computationally efficient least significant bit (LSB) data hiding is applied to the color encoded equirectangular projection of the sphere. As viewers put more attention to the areas around the equator of a 360-degree video compared to the poles, LSB data hiding may be performed in the regions around the poles without causing perceptually significant quality degradation. In addition, the equirectangular projection induces a huge amount of data redundancy in the areas near the poles which increases the capacity available for LSB data hiding. A performance assessment of the proposed spherical light-weight data hiding technique is conducted using the weighted-to-spherically-uniform peak-signal-to-noise ratio (WS-PSNR) and the Craster parabolic projection PSNR (CPP-PSNR). Because both metrics take the non-linear relationship between samples on the sphere and the samples mapped to the plane into account, they are well suited for assessing the performance of the spherical LSB data hiding technique. Numerical results are provided for scenarios in which a 360-degree cover video carries a secret video using different numbers of bit planes. It is shown that video fidelity in terms of WS-PSNR and CPP-PSNR is indeed kept high in the 360-degree stego-video as long as the LSB data hiding is performed in the areas around the poles. © 2019 IEEE.

The aim of steganography is to conceal the presence of communication by way of hiding secret data in perceptually irrelevant parts of a cover object. In this paper, we propose a method for hiding secret images in edge regions of high-definition (HD) images because the human visual system is less sensitive to intensity changes in these regions. In particular, least significant bit substitution is used to embed a secret image in the edge regions of a HD cover image. The edge regions are obtained using a Canny edge detector followed by morphological operations which are used to control the hiding capacity. A performance assessment of the proposed method reveals the trade-off between capacity, detectability, and perceptibility of the hidden data. © 2019 IEEE.

In this paper, we study moderately significant bit data hiding techniques for high-definition (HD) images. In contrast to least significant bit data hiding, we explore the potential of HD images to engage higher order bits for increasing the capacity of hiding secret images. In particular, the number of secret images embedded in moderately significant bits of a given cover image is successively increased to study the impact of this data hiding approach on image fidelity and image quality in the context of HD images. A comprehensive performance assessment is conducted on the stego-image carrying the secret images in terms of peak-signal-to-noise ratio (PSNR) and structural similarity (SSIM) index. It is shown that HD images indeed offer the potential of hiding several images within certain image fidelity and objective perceptual image quality constraints of the resulting stego-HD images. © 2018 IEEE.

Mobile services have seen a shift from voice services toward visual stimuli-based services ranging from mobile imaging over mobile gaming to upcoming mobile extended reality applications. In this paper, given the increased resolutions of the related mobile multimedia formats, we propose and examine positioning strategies for light-weight data hiding of secret images in moderately significant bits of high-definition (HD) cover images. Apart from linear and random positioning, visual attention mechanisms of the human visual system are addressed by separately utilizing either the background or center of the HD cover image for data hiding. A performance assessment of these positioning strategies is conducted in terms of the peak-signal-to-noise ratio (PSNR), structural similarity (SSIM) index, and visual information fidelity (VIF). It is shown that HD cover images indeed can carry more than a single secret image until noticeable quality loss is observed. Further, it is revealed that linear positioning of secret images in the whole, background, or center of HD cover images outperforms random positioning. As for the utilized performance metrics, it has been observed that PSNR cannot differentiate among the positioning strategies as it measures pixel-by-pixel differences and hence removes the impact of the position of the hidden data. As for the two quality metrics, compared to the SSIM index, VIF is found to stronger differentiate among the performance of the considered positioning strategies.

In this paper, we propose a viewing direction based least significant bit (LSB) data hiding method for 360◦ videos. The distributions of viewing direction frequency for latitude and longitude are used to control the amount of secret data to be hidden at the latitude, longitude, or both latitude and longitude of 360◦ videos. Normalized Gaussian mixture models mimicking the viewing behavior of humans are formulated to define data hiding weight functions for latitude, longitude, and both latitude and longitude. On this basis, analytical expressions for the capacity offered by the proposed method to hide secret data in 360◦ cover videos are derived. Numerical results for the capacity using different numbers of bit planes and popular 360◦ video resolutions for data hiding are provided. The fidelity of the proposed method is assessed in terms of the peak signal-to-noise ratio (PSNR), weighted-to-spherically uniform PSNR (WS-PSNR), and non-content-based perceptual PSNR (NCP-PSNR). The experimental results illustrate that NCP-PSNR returns the highest fidelity because it gives lower weights to the impact of LSB data hiding on fidelity outside the front regions near the equator. The visual quality of the proposed method as perceived by humans is assessed using the structural similarity (SSIM) index and the non-content-based perceptual SSIM (NCP-SSIM) index. The experimental results show that both SSIM-based metrics are able to account for the spatial perceptual information of different scenes while the PSNR-based fidelity metrics cannot exploit this information. Furthermore, NCP-SSIM reflects much better the impact of the proposed method on visual quality with respect to viewing directions compared to SSIM. © 2021 by the authors. Licensee MDPI, Basel, Switzerland.

In this paper, we propose a viewing direction weighted bit plane least significant bit (LSB) data hiding method for 360° videos. Normalized Gaussian mixture models (GMMs) of the distributions of the relative viewing direction frequency for latitude, longitude, and both latitude and longitude are used to describe the human viewing behavior of 360° videos on head-mounted displays (HMDs). Apart from formulating data hiding weight functions to control the amount of secret data to be hidden at given latitudes and longitudes, the GMMs are used in this work also to define viewing direction weighted bit plane functions. These viewing direction weighted bit plane functions specify the number of bit planes to be used for data hiding depending on the latitude and longitude. As humans pay more attention to the central front view of 360° videos when watched on HMDs, the number of engaged bit planes is decreased accordingly from the poles to the equator, and from the west and the east to the front view. A performance assessment of the proposed viewing direction weighted bit plane LSB data hiding method is conducted, showing the trade-offs between capacity and objective perceptual video quality. The results indicate that it is preferable to first utilize a given bit plane for data hiding in an entire 360° video frame before engaging the next higher bit plane subject to the viewing direction weighted bit plane functions. © 2021 IEEE.

In this paper, we propose a visual attention based least significant bit (LSB) data hiding method for 360° videos. In particular, information about the relative frequency of pixel access is used to control the amount of secret data to be hidden at the different latitudes of 360° videos. In this way, the typical behavior of users paying more attention to the equator region compared to the north and south poles when viewing 360° videos on a head-mounted display is taken into account. An analytical expression for capacity offered by this method to hide secret data in 360° cover videos is derived. A performance assessment of the visual attention based LSB data hiding method is conducted in terms of the peak signal-to-noise ratio (PSNR) and weighted-to-spherically-uniform PSNR (WS-PSNR) which both quantify the fidelity of a 360° stego-video with reference to the related 360° cover video. © 2020 IEEE.

Steganography has been widely used to embed secret data in cover multimedia carriers including digital image, audio, and video files such that the confidential information being communicated is hidden. Immersive multimedia, e.g., 360° videos, have recently received increased attention as promising applications for the Internet and 5G mobile networks. In this paper, we focus on using steganography for hiding secret data in an immersive multimedia carrier. Specifically, a spherical least significant bit (LSB) data hiding approach is proposed to embed secret videos in 360° cover videos along with morphological operations. Due to viewers are paying less attention to the North and South Poles compared to the Equator region of 360° videos, LSB data hiding is performed around the poles of the equirectangular projection of the sphere to the plane. Data hiding capacity is increased by using morphological operations in the remaining mid-region around the Equator which broadens the edge area for embedding secret data. The proposed spherical LSB data hiding technique may be used to hide secret videos or other confidential data in 360° cover videos. It can also be used to increase available bandwidth for communication purposes as long as embedded data has no significant impact on the quality as perceived by the end-user. A performance assessment of the proposed approach is conducted with respect to data hiding capacity and objective perceptual video quality. Numerical results are provided for a wide range of scenarios illustrating the trade-offs between capacity and quality. © 2019 IEEE.

Data hiding or information hiding is a prominent class of information security that aims at reliably conveying secret data embedded in a cover object over a covert channel. Digital media such as audio, image, video, and three-dimensional (3D) media can act as cover objects to carry such secret data. Digital media security has acquired tremendous significance in recent years and will be even more important with the development and delivery of new digital media over digital communication networks. In particular, least significant bit (LSB) data hiding is easy to implement and to combine with other hiding techniques, offers high embedding capacity for data, can resist steganalysis and several types of attacks, and is well suited for real-time applications. This article provides a comprehensive survey on LSB data hiding in digital media. The fundamental concepts and terminologies used in data hiding are reviewed along with a general data hiding model. The five attributes of data hiding, i.e., capacity, imperceptibility, robustness, detectability, and security, and the related performance metrics used in this survey to compare the characteristics of the different LSB data hiding methods are discussed. Given the classification of data hiding methods with respect to audio, image, video, and 3D media, a comprehensive survey of LSB data hiding for each of these four digital media is provided. In particular, landmark studies, state-of-the-art approaches, and applications of LSB data hiding are described for each of the four digital media. Their performance is compared with respect to the data hiding attributes which illustrates benefits and drawbacks of the reviewed LSB data hiding methods. The article concludes with summarizing main findings and suggesting directions for future research. This survey will be helpful for researchers and practitioners to keep abreast about the potential of LSB data hiding in digital media and to develop novel applications based on suitable performance trade-offs between data hiding attributes.