Abstract: In this paper, a vigorous watermarking system is proposed by using SVD(Singular Value Decomposition)and DWT(Discrete Wavelet Transform). SVD is being used for numerous applications along with supplementary methods In addition to usual hiding arrangements a random moniker is utilized to increase its heftiness in contradiction to unnecessary burglars. The unitary matrices are utilized to produce a moniker which is going to be embedded into the fourth level decomposition of shield image. After extraction, an image is checked with the Moniker embedded. If these Monikers are harmonized the unitary matrices will be used to excerpt watermark from the watermarked image.Diverse attacks are well-thought-out and the simulation outcomes show that the extraction of watermark after attacks had minor effect only.
Keywords: attacks, authentication, moniker, SVD, watermarking
The security of data transmitted from one place to other using regular wired or wireless network is a major concern in the current era. A large amount of secret information is being sent using a regular network it may be wired or wireless. Most of the networks will not employ additional security measures on how the data is being transferred from one place to other. Hence the users of these kinds of networks are themselves responsible to take care of their data. An attempt to transmit images safely was made in this paper. The components of a digital watermarking system are embedding and extraction. Depending on what domain of shield image was used to embed the watermark, spatial and transform or frequency domain watermarking systems are proposed in the literature.
The main advantage of transform domain tactics is their extreme robustness to common image falsifications. Discrete cosine transform is a fundamental technique which was used extensively in early days of image processing and specifically used in image compression. Later, the revolution of wavelet has shifted the view of a researcher as well as the people from industry with a wide range of applications. While the DCT provides an efficient representation for the input image, the DWT in addition to efficient representation it also enables different mechanisms by which different image processing tasks can ease their implementations and improves the performance of their task. The interesting feature of the wavelet transform is that it separates the image in terms of the frequency with multiple levels. Some of the levels are playing a vital role in reconstruction and other a minor role. Now the secret information can be saved in bands which has not much work in reconstruction.
This is somehow or other similar to the LSB based technique of spatial domain 1. The LSB based technique concentrates on hiding the message bits in the LSB locations of the cover image. The original LSB bits will be lost and the importance of these bits is less in the sense that these bits if changed from zero to one or one to zero results in a change of pixel value by only one, hence has less effect on the display. In addition to the said watermarking scheme, numerous watermarking techniques robust to symmetrical attacks have been proposed in the literature 23. The wavelet-based watermarking schemes are found to be robust against multiple attacks like compression, blurring, salt and pepper noise and many other 4.
A watermarking scheme will have the following basic components. Watermark is the prime component of the scheme, which is to be conveyed securely to the destination which sustains many attacks. Cover image, which is usually large enough for the watermarking scheme to embed the watermark on to it. Watermarked image, which is the outcome of the watermarking scheme. Embedding means a process of hiding the watermark in the cover image. An attack is an act of modifying the effective appearance or the effective pixel value plane to a different set. The extraction is the process of separating the watermark from the watermarked image 5.
In the literature, a number modifications and improvements have been made to the watermarking schemes. In 6, Mohammad Ali et. al presented a blind digital watermarking scheme based on quantization of Eigenvalues in Wavelet domain. In the literature fuzzy and artificial neural networks based techniques 7-9, SVD based techniques 1011, hybrid techniques 1213, Biometric template based techniques 14, Evolutionary algorithm based techniques 15, Quadtrees based techniques 16, GEP based techniques 17 and video watermarking schemes 1819 are proposed. A large number of surveys are also being conducted 2021.
In this paper a watermarking scheme is proposed which is a modified version or improved version of traditional SVD-DWT based watermarking scheme. The rest of the paper is organized as follows. In section-II, the basic or standard DWT-SVD based watermarking scheme was presented. In section-III the authentication issue of the standard DWT-SVD based technique is described. Section-IV is concerned the solution of the authentication problem. Section-V presents the simulation results and the last section concludes the paper.
II. The Standard DWT-SVD Watermarking Algorithm
The standard DWT-SVD watermarking scheme considers the cascade of DWT and SVD as the main building block for the watermarking scheme. The DWT will be applied to the cover image or the carrier image. The DWT usually decomposes the cover image into four frequency bands: Low-Low, Low-High, High-Low, and High-High. The Low-Low band characterizes low frequency, High-Low, and Low-High bands describe the middle frequency and High-High band characterizes high-frequency bands, respectively. The Low-Low band signifies approximate details, High-Low band horizontal details, Low-High vertical details and High-High band diagonal details of the image.
These different bands contain different grades of information in it. The cover image is basically an image hence the Low-Low band contains all the small variations which crucial in determining the boundary of different objects present in the image. Similarly, the Low-High and High-Low has some useful information in reconstruction than that of a High-High band of frequency. Hence the High-High frequency band was selected as the candidate to store the information related to the secret data. To provide additional security the SVD is applied to the High-High band of cover image as well as to the watermark. The concept of this scheme is to replace the singular values obtained after applying the SVD of cover image with the singular values of the watermark. It is observed that singular values range from85to 175 for most of the standard images.