Synchronisation correction-based colour image watermarking using proximal classifier with consistency and multivariate generalised Gaussian model

• Published in: Electronics letters Vol. 55; no. 24; pp. 1305 - 1307
• Main Authors: Wang, Qian, Wen, Taotao, Yang, Hongying, Wang, Xiangyang
• Format: Journal Article
• Language: English
• Published: The Institution of Engineering and Technology 28.11.2019
• Subjects: binary watermark image; colour images; common image processing attacks; corrected watermarked image; feature extraction; Gaussian distribution; geometric distortion parameters; geometric distortions; image coding; image colour analysis; image watermarking; multivariate generalised Gaussian distribution; multivariate generalised Gaussian model; proximal classifier; Signal processing; synchronisation; synchronisation correction‐based colour image watermarking approach; transforms; watermarking; YCbCr colour host image; geometric distortions; binary watermark image; common image processing attacks; image watermarking; Gaussian distribution; transforms; YCbCr colour host image; geometric distortion parameters; multivariate generalised Gaussian distribution; synchronisation; watermarking; feature extraction; synchronisation correction-based colour image watermarking approach; colour images; multivariate generalised Gaussian model; image colour analysis; proximal classifier; image coding; corrected watermarked image
• ISSN: 0013-5194; 1350-911X; 1350-911X
• DOI: 10.1049/el.2019.2586

In this Letter, a novel synchronisation correction-based colour image watermarking approach is proposed to resist geometric distortions. The Cb channel of the YCbCr colour host image is first decomposed by a two-level contourlet transform, and binary watermark image is embedded into the host image by modifying the lowpass coefficients. Then the multivariate generalised Gaussian distribution is used to capture the significant features of colour images, and the proximal classifier with consistency is employed to estimate geometric distortion parameters. Finally, the watermark image is extracted from the corrected watermarked image. Simulation results demonstrate that the presented approach shows excellent robustness against various common image processing attacks and geometric attacks while guaranteeing invisibility.