Abstract: As to the digital watermarking based on local feature is lack of robustness to non-scaling attacks,this paper put forward a watermarking algorithm based on template and local feature. Firstly, the template information was used to correct the non-scaling attacks. Then the Harris-Laplacian detector was utilized to extract steady feature points from the host image, and the local feature regions area were determined adaptively according to the feature scale theory; finally, the digital watermarks were embedded into the local feature regions using cirque quantization. Experimental results show that the proposed scheme is not only invisible and robust against common signals processing, but also robust against the geometric attacks, in particular,the non-scaling.

Key words: geometric attack, Harris-Laplacian detector, template, non-scaling, cirque quantization

摘要： 针对目前基于图像特征的数字水印对非等比例缩放攻击鲁棒性不足的现状，提出了一种模板和图像特征相结合的水印算法。该算法首先利用模板信息纠正非等比例缩放的攻击，再用HarrisLaplacian检测算子从载体图像中提取出稳定的特征点,然后根据特征尺度自适应确定局部特征区域,并采用圆环量化的方法将水印信息嵌入到局部特征区域中。实验表明,新算法可以保证水印的不可见性,而且对常规信号处理和几何攻击尤其是非等比例缩放具有较好的鲁棒性。

关键词: 几何攻击, Harris-Laplacian检测算子, 模板, 非等比例缩放, 圆环量化