Exploring DWT–SVD–DCT feature parameters for robust multiple watermarking against JPEG and JPEG2000 compression

Abstract

This paper presents a novel scheme to implement blind image watermarking based on the feature parameters extracted from a composite domain including the discrete wavelet transform (DWT), singular value decomposition (SVD), and discrete cosine transform (DCT). Multiple bits can be embedded into a single image block by adjusting designated parameters via a progressive quantization index modulation technique. The quantization with respect to the feature parameters obtained in the DWT–SVD–DCT domain leads to efficient watermark extraction without referring to the original image. Experimental results show that the embedded watermarks exhibit exceptional robustness against image compression using JPEG and JPEG2000 coding standards.