基于两级内容表征的数字图像内容保护算法

文中首先通过动态加权平均得到图像的象素级内容表征(Pixel-level Image Content Map，PICM)，在此基础上，通过图像分割技术得到图像的区域级内容表征(Regional Image Content Map，RICM)，并将表征图像RICM的Hu不变矩编码为内容7F-gF45息嵌入到原图像中，从而实现对原始图像的内容保护。实验结果表明，由于该两级内容表征方法及其矩特征的良好性能，算法对图像的非内容级图像操作(特别是仿射变换)具有很强的鲁棒性，且能够敏感的检测出对所保护图像的内容篡改。

Image Content Protection Based on 2-Level Content Representation

A novel image content protection(ICP) method is proposed in this paper. To satisfy the fundamental requirement for ICP, i.e. being robust to non content operations and sensitive to content tampers simultaneously, a pixel level image content map(PICM) is extracted by dynamic weighted averaging. Then, a regional image content map(RICM) is derived through segmentation of PICM and its invariant moments are converted to watermarking bit stream that is embedded in the source image robustly to achieve protection of the content. Various experiments show that the 2 level content representation, i.e. PICM and RICM, and its invariant moments have excellent performance in ICP, and this method meets the fundamental criterion well.