Dual domain robust watermarking scheme using random DFRFT and least significant bit technique

Abstract

This paper presents a novel image watermarking scheme that uses multiple-parameter discrete fractional Fourier transform (MPDFRFT) with random DFRFT and least significant bit (LSB) technique. The proposed scheme uses the LSB technique to embed watermark image in an original image, while RDFRFT and MPDFRFT are used to enhance the security of the watermarked image. The presented method aims to make the system more robust and secure at two different levels along with faster embedding and extraction process. The MPDFRFT is input data points of order DFRFT parameters. The MPDFRFT can be converted to the DFRFT when all of its order parameters have same values. On the other hand, RDFRFT is kernel matrix with random values of their DFT eigenvectors and Eigenvalues. Random magnitude and phase of its transformed output are unique features of RDFRFT that utilised to enhance security. LSB technique is preferred for watermark embedding because of its lesser effect on the image perceptibility. In proposed scheme, the cover image is converted into sub-images and MPDFRFT is applied to each sub-image. Then RDFRFT is applied in tandem to transformed sub-images in order to enhance robustness followed by watermark embedding using LSB technique. In watermark extraction process the inverse of MPDFRFT and RDFRFT is applied for reconstruction of original images along with LSB extraction mechanism. The performance of the proposed system is evaluated by using MATLAB software under various attacks and distortions. The simulation results validate that the embedding scheme presented in this work has better performance in terms of robustness along with perceptibility and security over contemporary existing schemes.