Watermark re-synchronization using sinusoidal signals in DT-CWT domain

Abstract

Embedding sinusoidal signals or tiles patterns into image in the spatial domain to form some peaks is an effective technique for geometric invariant image watermark detection. However, there are two drawbacks in these spatial domain based schemes: one is poor picture quality of resulting watermarked image and the other is weak peaks visibility which is hard to detect. The previous works suffer from requiring a very strong watermark embedding to ease re-synchronization peaks finding, which in turn leads to a poorly watermarked image. To overcome this problem, we explore embedding sinusoidal signals individually in each of the selected sub-bands of dual-tree complex wavelet transform domain (DT-CWT), and then detecting the re-synchronization peaks by using the accumulated embedding sub-bands. Experimental results demonstrate that by adopting our approach, the resulting re-synchronization peaks are robust to rotation, scaling and translation attacks while preserving the visual quality of the watermarked image, thereby resolve the unavoidable dilemma faced by the other schemes.