The limitations and applications of various state-of-the-art hybrid digital image watermarking methods are summarized in Table 5. These techniques increase higher computational complexity time and space. Also, they cannot achieve high performance with a trade-off between imperceptibility, robustness, security, and capacity.
Moreover, security is a big challenge for this technology. However, the recent induction of the Internet of Things IoT and blockchain-based authentication provide better security. IoT devices use lightweight encryption algorithms for enhancing security, whereas a computer uses sophisticated encryption algorithms. Besides, IoT needs limited storage and requires less computing complexity [ 35 ].
On the other side, a blockchain-based authentication is a decentralized approach. A blockchain contains digital information blocks that are linked together and secured by cryptography.
This blockchain technology stores the signature of the host image on the blockchain. Therefore, the host image cannot be modified by an attacker. For locating the tampered region and authenticity of the host image, the signature can be recovered from the blockchain. We have highlighted some of these techniques below. For checking the authenticity of pharmaceutical products in the IoT environment, a method is proposed that uses the NFC or near-field communication [ 37 ]. The method ensures session key security by using the ROR real-or-random model which is effective from computation and communication cost perspectives.
A content-based image retrieval CBIR method is proposed without accessing cloud-server information [ 38 ]. Here, images are represented by extracting feature vectors. The secure k-nearest neighbor kNN algorithm protects the feature vector. For ensuring the security of existing devices, a blockchain-based secured mutual authentication termed as BSeIn method is proposed that ensures privacy and security [ 39 ]. The method is secured against various attacks [ 40 ]. In addition, a new authentication method related to cloud-assisted cyber-physical system CPS is designed in reference [ 41 ].
Here, the external user can access the cloud server information. Cloud server data can communicate securely by the authentication scheme between a cloud server and a smart meter.
The method ensures the security of the system. Recently, Wazid et al. The method uses cryptographic hash functions and bitwise XOR operations and has shown better security experimentally. Current trends of image authentication are based on the frequency or transform domain, whereas some are based on the spatial domain.
But, now research trends are integrated with artificial intelligence. So, some new open research issues are growing. They are given as follows: i Voice to Authentication. It is the ultimate desire to have image authentication along with the voice. Technology needs to integrate the watermark text from the direct voice. At first, voice to text conversion will be done and then the text will be inserted to the cover image. It will be more secured than traditional authentication for the extra step voice to text as it finds applications in natural language processing.
At present, fake accounts can be seen on social media which harms society. Various crimes, fake news, and rumors are spreading on an increasing scale through these fake profiles.
So, to stop these unwanted profile activities special research focus is needed. This research will lead to the privacy-preserving image authentication system. Image authentication is a costly operation when we consider a big image database. So, image databases should be encrypted before sending it to the cloud.
Blockchain technology can be integrated here for better security. It is a challenging issue to find out the original watermark from the encrypted cover image on the receiver side and an open problem for future researchers. Images are an important part of multimedia data. Image authentication is a challenging task due to Internet traffic.
Because of the interactive communication of multimedia data and wide-spread use of IoT technology, information can be duplicated easily. Along with image data security, it is essential to ensure the imperceptibility, robustness, and enhanced data embedding capacity.
Hybrid digital image watermarking is a significant field for ensuring these issues. However, our investigation has found that the existing hybrid methods need to be improved to ensure these issues. Moreover, we have pointed out that the recent induction of the Internet of Things IoT and blockchain-based authentication provide better security.
Therefore, to get improved robustness and high image data security along with better imperceptibility and embedding capacity, future researchers must combine machine learning and artificial neural network algorithms in the hybrid transform domain.
This is an open access article distributed under the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Article of the Year Award: Outstanding research contributions of , as selected by our Chief Editors. Read the winning articles. Journal overview. Special Issues. Academic Editor: Patrick Seeling. Received 22 Feb Revised 21 Jun Accepted 21 Jul Published 06 Aug Abstract Digital image watermarking is an attractive research area since it protects the multimedia data from unauthorized access.
Introduction Multimedia technology is improving day by day. The contributions of this research are i We present the current trends of hybrid methods ii We identify the limitations of the state-of-the-art hybrid methods of digital image watermarking iii We point out the challenges that must be addressed by future researchers In this research, the existing hybrid domain-based digital image watermarking methods are reviewed in Section 2.
Literature Review In recent years, many hybrid digital image watermarking methods have been developed for hiding data to take into account increasing robustness, capacity, and security by maintaining the visual quality of the watermarked image. Framework of Hybrid Methods of Digital Image Watermarking Digital watermarking technology is the way of altering multimedia data by adding information into the host media to protect its copyright information [ 27 ].
Figure 1. Figure 2. Framework of hybrid methods of digital image watermark embedding. Figure 3. Figure 4. Table 1. Table 2. Table 3. Table 4. Table 5. Figure 5. References P. Uno, M. Kawato, and R. Feng, L. Zheng, and P. Zhou, J. Ma, and W. Hemdan, N. El-Fishaw, G. Attiya, and F. Tun and Y. Jane, E. Sharma, A. Singh, and S. Al-maweri, W. Adnan, A. Rahman Ramli, K. Samsudin, and S. Hu and L. Natu, P. Natu, and T. Assini, A.
Badri, A. Badri, K. Safi, A. Sahel, and A. Takore, P. Kumar, and G. Jain and U. Hamidi, M. Haziti, H. Cherifi, and M. Zhang, Y. Frequency domain capabilities and mixed-domain techniques, when added to signals, are believed to provide the right amount of robustness in order to guard against watermark attacks. The publisher Playboy has used an invisible form of digital watermarking to detect where its copyrighted material has been illegally posted on other websites.
By: Justin Stoltzfus Contributor, Reviewer. By: Satish Balakrishnan. Dictionary Dictionary Term of the Day. Natural Language Processing. Techopedia Terms. Connect with us. Sign up. Term of the Day. In section , I discuss algorithm, application and attack in digital watermarking, in all these three sections, I come up with two points: traditional and novel.
In section 6, I try to assume some directions of developments of digital watermarking. In this section, I use three subsections to briefly introduce some basic knowledge of digital watermarking from three aspects which may be useful in the follow sections. Visible and invisible are the two basic types of digital watermarking, and every digital watermark can be considered as either visible or invisible.
Visible digital watermarking is a way by which anybody can put visible information in digital signal, the information is often a logo, which identifies the owner of the digital signal.
For example, a television broadcaster usually adds its logo to the corner of its video, this is a typically visible digital watermark. Invisible digital watermarking is a way by which anybody can hide information in digital signal and the information will not be perceived.
Since it is invisible, invisible digital watermarking has a widespread use. It can be used to add identification of owner in signal and is more difficult to detect and remove.
It is also possible to use embedded information to share secret or communicate in a hidden way. The invisible digital watermarking can be detected and validated by some specific technology or the people share secret with the owner.
The research on digital watermarking algorithm, application and attack are most about invisible digital watermarking. Generally, a digital watermark can be embedded into all forms of media. The most common medium are audio, video and picture. It is easy to add a visible digital watermark on a digital signal, it just needs to add some data on original signal.
But to make an invisible digital is not so easy as visible digital watermarking. Different medium has different data structure, so according to different medium, various algorithms are used to add digital watermarks in signal without changing the way which original signal looks like. There are various ways to classify digital watermarking, such as: by feature, by medium, by detecting process, by content, and so on.
Below are some useful and effective ways:. Robustness is one of the most important attributes of a digital watermark. A fragile digital watermark is a digital watermark that fails to be detected after the slightest modifies. A semi-fragile digital watermark is a digital watermark that resists benign transformation but fails to be detected after malignanttransformations. A robust digital watermark is a digital watermark that resists a designated class of transformations.
It does not mean that a robust digital watermark is better than a fragile digital watermark. Fragile and semi-fragile digital watermarks are commonly used to detect malignant transformations and protect the integrity of the digital signal. Robust digital watermarks are often used in copy protection applications. It is a way to determines two different main classes of digital watermarking schemes by the length of the embedded message.
In zero-bit or presence watermarking schemes, the message is conceptually zero-bit long, it is designed to detect the presence or the absence of the digital watermark in the marked object. In multiple-bit watermarking or non-zero-bit watermarking schemes, the n-bit-long stream message is modulated in the watermark. If a digital watermarking requires the original data for watermark, it is called non-blind watermarking. If a digital does not require the original data for watermark, it is called blind watermarking.
If the marked signal is obtained by an additive modification, this kind embedding method is called spread-spectrum. If the marked signal is obtained by quantization, this kind embedding method is called quantization type.
If the marked signal is embedded by additive modification in the spatial domain, this kind embedding method is called amplitude modulation.
Spread-spectrum digital watermarks have the best robustness but weak in capacity. Quantization digital watermarks are known to be weak in robustness but have great capacity. Algorithm is the core of digital watermarking, a good algorithm can make the digital watermarking more robust and usable.
In this section, paper will introduce spatial domain and algorithms based on it. In the early days, the digital watermarking algorithms are mainly based on spatial domain. To gray-scale images, every pixel is 8-bit, and by the Most Significant Bit begins to right Least Significant Bit, which implies the importance of data bits order.
Hence, we can embed the watermarks by editing the Least Significant Bit. It is an easy and basic way in digital watermarking. But compared with the transform domain algorithms which will be traduced below, algorithms based on spatial domain are fragile. The DWT Discrete Wavelet Transform is a powerful and useful multi-resolution decomposition method in digital watermarking. It is often applied on image processing, and has been applied to such as noise reduction, edge detection, and data compression.
It is consistent with the visual perception process of human eyes. DWT can localize the signal in spatiotemporal, it is a new signal analytic theory but has already been widely used. DWT uses discrete wavelet transform to decompose the original image into four sub-bands LL1, HL1, LH1, and HH1, which can be separate into lower frequency sub-bands and higher frequency sub-bands. We can reach the final satisfied scale by repeat this decomposition process.
The low frequency image usually has better stability against the image distortion, so most time digital watermarking based on DWT is done in the LL sub-band to be robust to various classes of attacks like filtering, collusion and compression. DWT is easy to implement and can efficiently reduce the computation time. The x n in the Fig. The g n is the low pass filter and the h n stands for the high pass filter.
A single level of decomposition can be expressed as in Eq. The decimation and filtering process is continued until we reach the desired level which depends on the length of signal. Then we concatenate all the coefficients, start from the last level of decomposition to get the DWT of the original signal.
0コメント