一种基于ICA的脊波域数字水印算法

An ICA-Based Digital Watermarking Algorithm In the Ridgelet Transform Domain

鲁棒性是数字水印的基本要求之一,水印嵌入的位置和强度直接影响到水印的鲁棒性。脊波变换是新型的多尺度分析方法。本文提出的脊波变换是图像在有限Radon变换(FRAT)的基础上进行二维小波变换,将Arnold置乱后的水印嵌入到图像的脊波系数中,利用独立成分分析(ICA)进行水印提取。实验结果表明,在强烈攻击的情况下,此方法具有良好的水印不可见性,对Gaussian、JPEG、Speckle攻击具有良好的鲁棒性。

Robustness is one of the most basic requirements for digital watermarks. The embedding position and embedding strength will affect the robustness of watermarks. Ridgelet Transform is a new multi-scale analysis. This paper presents the ridge of wavelet transform based on limited Radon transform images （FRAT）, and a two-dimensional wavelet transform. Arnold Scrambling is adopted in the embedding of image ridgelet coefficients, and independent component analysis （ICA） is used for watermark extraction. The experimental results show that this method have good no-visibility of the watermarks, and is robust to the Gaussian, JPEG, Speckle attacks.