Watermarking of EEG Data to Provide Security Based on DWT-SVD and Optimized by Firefly Algorithm

Introduction

Nowadays, most of the data is accessible in computerized form on the web. This accessibility of information on the web permits users to share and access all the information and data digitally, which encroaches on the law of copyright possession. The data embedded with a watermark can be utilized for sharing between the sender and an authenticated receiver only. Digital watermarking permits the user to add information like a logo or image to a message to demonstrate ownership, authenticity, and duplicity control. The process of watermarking has different applications, like copyright security, fingerprinting, information authentication, and a lot more (Garg and Kishore, 2020). The watermarking algorithm must keep up with the nature of the original image along with watermark embedding that highlights imperceptibility and robustness against different attacks like rotation, scaling, translation, etc. (Priya et al., 2017).

In this paper, the watermarking technique DWT-SVD-FA is proposed to perform secured watermarking over EEG data so that the privacy of patients and their information is maintained. Here, the firefly algorithm has been utilized to optimize the scaling factor and the watermark logo was successfully implanted and extracted at the time of requirement. Watermarking of digital information can be performed either in the transform or in the spatial domain. Pixel values of an image can directly be utilized in the spatial domain, whereas they need to be converted into frequency coefficients at the time of transform domain watermarking. In the spatial domain, performing watermarking is easier, but efficiency can be lower (Kazemivash and Moghaddam, 2017). To perform watermarking in the transform domain, there are a number of methods like DWT, DFT, DCT, RDWT, LWT etc. These schemes provide higher efficiency and flexibility, but we need to be careful as watermarking in lower frequency components may change the original image, whereas in the high frequency region it may affect during noise attack like low pass filtering by which the watermark shows floppiness. There is a huge amount of research going on in the wavelet domain, as it shows higher efficiency and faster response as compared to other available techniques.

For watermarking, a suitable sub frequency band can be selected in the DWT methodology, which makes a minor change to the watermarked image. Agoyi (2020) has proposed a DWT using chrip z-transform and SVD for the purpose of image watermarking and shows good robustness against various attacks and imperceptibility. Luo et al. (2020) have proposed a multi-scale and protected image watermarking methodology which works on the basis of Integer wavelet transform (IWT) and SVD. Here, the author has performed first level IWT on image data to obtain four sub bands and, further, corresponding to each sub band, diagonal singular matrices have been evaluated using SVD. To embed a watermark, each matrix is divided into four parts to match the size of the watermark. Here, the watermark is directly embedded into the message with the use of various scaling factors. The author has also utilized optimization techniques based on three-dimensional optimality, to improve robustness and imperceptibility. For authentication and copyright protection, a robust watermarking methodology has been proposed by (Liu et al., 2019). Here, the author has worked based on DWT, Hessenberg decomposition (HD) and SVD. Using multi-level DWT, a number of sub-bands have been obtained with the decomposition of image data. Using HD, coefficients have been evaluated along with SVD decomposition of the watermark. Furthermore, the embed watermark scaling factor has been optimized through natural optimization method. With the use of a fruit fly optimization algorithm, the results indicate invisibility and robustness. To solve issues of security, a robust watermarking method has been proposed by (Amrit et al., 2021) that utilizes DWT-SVD with an optimization algorithm has been proposed. To improve the performance capacity in terms of visual quality and robustness, the author uses a dual watermarking scheme over medical images. The author has achieved more robustness, imperceptible and good capacity simultaneously.