Robust image watermarking based on hybrid transformation

Abstract

Watermarking techniques are mainly used for protecting intellectual property right. The transparency and robustness of watermarks are critical. However, they are often conflicting. We propose a new image watermarking scheme, which incorporates two kinds of transformation technologies to provide the better compromises between the visual quality and the robustness for resisting various signal processing or degradation. This algorithm utilizes the wavelet multiresolutional structure to select the location for the watermark. The quantization errors in the discrete cosine transform domain achieve the smallest visual degradation. Embedding the watermark within the image by introducing reduced distortion into transformed coefficients provides the robustness for the watermark detection without degrading the image quality. Besides, the proposed method does not require the original image to detect the watermark. Some experimental results demonstrate that the proposed scheme is more robust than previous schemes in terms of certain perceptual quality.