Abstract: This paper proposes a novel feature orientation watermark algorithm based on digital wavelet transform (DWT). Feature points are extracted by utilizing the characteristic of image from the low frequency sub-band of wavelet coefficients as index of watermark embedding positions. The technology is suitable to watermark blind detection. Simulation results show that algorithm has prefect positioning accuracy and unaffected characteristics. Practicality is perfect due to good balance between transparency and robustness.

Key words: Digital watermark, Wavelet transform, Feature orientation

摘要： 提出了一种基于小波变换的特征定位数字水印算法，利用图像自身特性，在小波系数低频子带中提取特征点作为水印嵌入位置的索引。该技术适合于水印盲检测，仿真表明，算法在检测时定位精度很高，基本不受水印嵌入的影响，具有很好的透明性和鲁棒性，实用性也较高。

关键词: 数字水印, 小波变换, 特征定位