A Novel Semi-blinding Watermarking Algorithm in the Wavelet Domain

A new semi-blinding watermarking algorithm in wavelet domain is proposed in this paper. The original image is first decomposed in wavelet domain, then the feature points are extracted from the lowest approximation with the scale interaction model for feature detection and the image feature tree in wavelet domain is created corresponding to the feature points. The binary watermark bits are embedded in all the nodes of the feature tree. Watermark extraction only needs the feature points position but not the original image. Since the watermark bits are inserted into the features of image, it is robust to common image processing, especially to the lossy compression such as JPEG, JPEG2000 and Gaussian noise. It belongs to the second generation watermarking schemes.