Investigation on improved watermarking through maximum entropy and normalized principal components in dwt domain

Abstract

In the era of multimedia networks, multimedia contents that include text, audio, image and video information are widely circulated among various end users. When the multimedia contents are transferred or stored in public networks/domain, the protective mechanism should be in place to ensure the safety of the images from malicious access, image forgery, copyright and ownership violations. These issues attract researchers to provide effective solutions in the form of watermarking, authentication and cryptography algorithms. Watermarking is the process of embedding a secret image into a host image without degrading the imperceptibility of the observer. Digital watermarking is a type of data hiding tool which embeds a watermark into a cover image thereby providing solutions to ownership and tamper localization issues.In recent years, semi-fragile watermarking has gained momentum and attracts the researchers for the employment of frequency domain techniques. Spatial domain watermarking schemes need less computation but often result in poor robustness against geometric attacks. Frequency domain algorithms provide an alternative for this issue and hence the watermarking schemes based on Discrete Cosine Transform (DCT), Discrete Wavelet Transform (DWT), Redundant Discrete Wavelet Transform (RDWT) and Singular Value Decomposition (SVD) have gained interest in watermarking concepts. SVD combined with the transforms such as DCT, DWT, RDWT are more efficient than SVD based methods. DWT-SVD based watermarking schemes provide better robustness and imperceptibility due to the sensitiveness to the modeling of DWT. In DWT-SVD watermarking schemes, watermark (WM) embedding is carried out on S matrices of LL subbands of the host and the WM images. newline