Skip to Main Content
A new image adaptive watermarking algorithm based on Bidimension Empirical Mode Decomposition (BEMD) is proposed. As a new two-dimensional signal processing tool, BEMD is first introduced into the image watermarking field in this paper. In the proposed algorithm, the cover image is decomposed into several intrinsic mode functions (IMFs) based on its local feature adaptively. Then the human visual system (HVS) model function based on IMFs and the logo image is generated, and inserted into the cover image. BEMD technique is used in the extracting processes as well and the recovered watermark image is obtained clearly. We have tested the robustness of the proposed scheme under cropping, JPEG compression, resizing, median filtering, and additive white Gaussian noise (AWGN) attacks. The experiments indicate that this technique can achieve high imperceptibility while showing good robustness. Furthermore, the proposed method is compared with the block DCT based method, and the experiment result confirms that the proposed method shows better performance.
Date of Conference: 11-13 July 2007