Robust and Secure Watermarking Scheme for Multimedia Data against Various Attacks

Abstract

The widespread adoption of digital technology has become prevalent in various domains. Concurrently, the internet began reshaping the world into a closely interconnected global community. While digital data offered numerous advantages over analog inputs, including ease of storage and transmission, many providers of multimedia content expressed reservations. They were concerned that the high-quality replication made possible by digital files might infringe upon their copyrights.Since necessity is the creator of innovation, watermarking methods were developed to address this issue.A hidden and undetectable piece of data is inserted into the digital file during watermarking.Such confidential information might be a logo or another piece of information that helps users identify the content source or creator. By obtaining concealed information from the digital data, the source owner might readily establish his ownership in the event of a dispute.The current work focuses on imaging watermarking to provide the confirmed authenticity of transferred multimedia files. This is done in light of the abovementioned and recognizing the broad applications of multimedia as well as its associated security issues. The thesis proposes several enhanced methods of multimedia image watermarking that give superior robustness and show the positive quality, greater embedding capacity, and confidentiality of the watermark after evaluating the present techniques for multimedia image watermarking. In this thesis, four kinds of problems have been proposed related to watermarking techniques for multimedia data. The first is concerned with a method for watermarking images utilizing the discrete cosine transform (DCT), bi-dimensional empirical mode decomposition (BEMD), and particle swarm optimization (PSO). DCT coefficient is utilized on the initial image during the encoding process, and then the watermark image is divided apart using BEMD decomposition. Particle swarm optimization (PSO) is employed to optimize complicated and multidimensional searches. The security key is used to embed the scaling and embedding factors. IDCT and IBEMD are used during the process, and a recovery procedure is also utilized to extract the watermark image. The second deals with introducing robust image watermarking using SVD (singular value decomposition), BEMD, DCT, PSO, andDWT (discrete wavelet transform). Second-level DWT is employed throughout the embedding procedure to deconstruct the cover image and watermark the image into sub-bands. Moreover, BEMD decomposition is implemented at the chosen DWT frequency. For optimization, PSO is employed for complicated as well as multidimensional searches. Moreover, DCT, as well as SVD, are implemented on the selected band. The embedding and the scaling factor are encoded with the aid of a security key. This technique is followed by ISVD, IDCT, IDWT, and IBEMD. The suggested method produces watermarked images with superior visual clarity and robustness against various attacks such as speckle, median filter, shearing, salt and pepper, gamma correction, scaling,Gaussian filter, and rotation for greyscale images. The third issue relates to an improved watermarking method using DWT, SVD, and BEMD where the watermark image is scrambled with Arnold transform to enhancethe algorithm's security. The BEMD disintegrates the input image in this instance to provide the multi-scale demonstration in the form of IMFs and residue. The suggested algorithm offers greater robustness and imperceptibility against several geometrical and non-geometric attacks. The last issue addressed in this study pertained to a watermarking method that combined DWT and BEMD for ownership verification, aiming to provide markedly improved visual quality compared to alternative decomposition techniques.BEMD can identify the image from the lowest fragile frequencies to the most robust frequencies. Also, the performance comparison with other competitive methodologies provides superior outcomes for various attacks.