Two dimensional time-frequency analysis based eigenvalue decomposition applied to image watermarking

• Published in: Multimedia tools and applications Vol. 49; no. 3; pp. 529 - 543
• Main Authors: Stanković, Srdjan, Orović, Irena, Žarić, Nikola, Ioana, Cornel
• Format: Journal Article
• Language: English
• Published: Boston Springer US 01-09-2010; Springer Nature B.V; Springer Verlag
• Subjects: Computer Communication Networks; Computer Science; Data Structures and Information Theory; Decomposition; Digital imaging; Digital rights management; Eigenvalues; Engineering Sciences; Fourier transforms; Frequency distribution; Image processing systems; Mathematical models; Multimedia Information Systems; Pixels; Signal and Image Processing; Special Purpose and Application-Based Systems; Studies; Time-frequency analysis; Two dimensional; Watermarking; Eigenvalue decomposition for image regions classification; Digital image watermarking; Two-dimensional time-frequency analysis
• ISSN: 1380-7501; 1573-7721
• DOI: 10.1007/s11042-009-0446-x

An application of two-dimensional time-frequency analysis and corresponding two-dimensional eigenvalue decomposition for image watermarking purpose is proposed. The eigenvalue decomposition is used to provide a criterion for watermarking coefficients selection. It is primarily used to select pixels suitable for watermarking that belong to busy image regions. The watermark embedding is performed in the space/spatial-frequency domain by using middle frequency components, whose number is determined from the eigenvalue decomposition, as well. In order to provide its imperceptibility, watermark is modelled and adapted to the local frequency content of each considered pixel. For an efficient watermark modelling procedure, the concept of space-varying filtering is employed. Furthermore, the watermark detection is done within the space/spatial-frequency domain, which facilitates detection process due to the larger number of coefficients comparing to the space or frequency domain, separately. The efficiency of the proposed procedure and its robustness in the presence of various attacks is proven on the examples.