中文摘要：

针对几何攻击所导致的水印不同步问题，提出了一种结合Zernike矩和图像归一化的有效视频对象水印算法。Zernike矩的幅度具有旋转不变性，缩放和平移不变性通过图像归一化取得。水印嵌入采用零水印方案，解决了基于Zernike矩的图像重构效果不理想和重构过程中复杂度高的问题。实验结果表明，该水印算法对旋转、缩放等几何操作具有鲁棒性，同时对压缩、滤波、高斯噪声等常见的图像处理操作也具有鲁棒性。

英文摘要：

In order to resist geometrical attacks, an object-based video watermarking algorithm using Zernike moments and image normalization is proposed in this paper. Rotation invariance is achieved by taking the magnitude of the Zernike moments. Image normalization method is used for scale and translation invariance. Though watermark signal can be embedded by modifying the Zernike moments of the cover image, the reconstruction procedure will suffer the fidelity degradation. Also, the reconstruction procedure requires Zernike moments with high order and this leads to a heavy computation load. In order to solve these problems, zero-watermarking technique is employed. Instead of embedding watermark, zero-watermarking technique extracts some essential characteristics from the host signal and uses them for watermark detection. Experimental results show that the scheme is robust against rotation, scaling, as well as a variety of common manipulations such as lossy compression, filtering and additive noise.