An Adaptive Image Watermarking Based on Wavelet Transform and Shift-Orthogonal Finite-Length Sequences

摘要

In recent years, digital watermarking has become the important method of copyright protection. This paper presents an adaptive image watermarking scheme embedding shift-orthogonal finite-length sequences into wavelet domain. The sequence has the characteristics of shift-orthogonality and mean 0, variance 1. Because the parameters used to generate the sequence are much, the security of this sequence is higher than pseudo random sequence. Watermark embedding is adaptive based on the characteristics of the human visual system (HVS). In the process of watermark detection the normalized correlation function is used. The experimental results show that the proposed watermarking scheme is perceptual invisible and robust against attacks such as JPEG compression-additive noise, filtering and cropping.