Research on Robust Digital Watermarking Resilient to Geometric Distortion

Abstract

With the rapid development of the Internet, digital multimedia information has become an indispensable part of people’s lives, and the copyright issues are becoming increasingly prominent. As an important method to solve the copyright problem, digital watermarking technology embeds a specific piece of information named watermark into the carrier information, which can provide proof of ownership for the copyright owner.

The robust blind watermarking based on uniform log-polar mapping (ULPM) is implemented, the details are improved, and the algorithm flow and parameter values are optimized. The watermark used for embedding consists of two parts: informative sequence and tracking sequence. The watermark embedding algorithm embeds the watermark into the DFT amplitude spectrum of the carrier image according to the ULPM mapping table. The watermark extraction algorithm locates the watermark by performing cross-correlation operation, and the message is extracted. The watermark embedding end and the watermark extraction end only need to share one key to realize the blind extraction of the watermark. In order to improve the invisibility, distortion compensation is performed in the spatial domain based on the masking characteristics of the human visual system after embedding the watermark, while error correction coding and spread spectrum technology are used to improve the robustness of the watermark. Based on these, a watermark comprehensive evaluation system is designed and implemented. Eight representative images are selected from the standard test images database as the test image set. The experimental results demonstrate that the proposed watermarking scheme has good invisibility and computational efficiency. And the watermark is robust to common signal processing such as JPEG compression with a quality factor of 20 or higher, Gaussian filtering with a filter size of 3×3, and so on. It is also robust to common geometrical attacks such as rotation at any angle, proportional scaling with a scaling factor of 0.75 to 1.40, and cropping with a ratio of more than 45%.

Key Words

information hiding; digital watermarking; geometric attacks; uniform logpolar mapping; distortion compensation

标题

抗几何失真的鲁棒数字水印的研究与实现

摘要

随着互联网的迅速发展，数字多媒体信息已成为人们生活中不可或缺的组成部分，随之而来的版权问题日益凸显。数字水印技术将一段特定的信息即水印嵌入到载体信息中，为版权所有者提供所有权证明，是解决版权问题的一种重要方法。

本课题实现了基于均匀对数极坐标映射 (ULPM) 的鲁棒盲水印框架，并完善 了其细节，优化了算法流程和参数取值。用于嵌入的水印由两部分组成：消息序 列和跟踪序列。水印嵌入算法根据 ULPM 映射表将水印嵌入到载体图像的 DFT 幅度谱中。水印提取算法通过互相关运算来定位水印，从而提取出其携带的消息。 水印嵌入端与水印提取端只需要共享一个密钥，就能实现水印的盲提取。在嵌入 水印后，基于人类视觉系统的掩蔽特性在空域进行失真补偿，以提升其不可见性； 使用纠错编码、扩频技术，以满足其鲁棒性。在此基础上，设计并实现了一个水 印综合评估系统，从标准测试图数据库中选取了 8 幅具有代表性的图像作为测试 图像集。实验结果表明，提出的水印方案具有良好的不可见性、计算效率。对质 量因数为 20 以上的 JPEG 压缩、滤波器大小为 3×3 的高斯滤波等简单攻击，任意角度的旋转、缩放因子为 0.75~1.40 的等比例缩放、比例为 45% 以上的裁剪等 几何攻击具有良好的鲁棒性。

关键词

信息隐藏；数字水印；几何攻击；均匀对数极坐标映射；失真补偿