【摘要】 研究一种改进的基于运动矢量的Mpeg2视频水印技术.利用Mpeg2标准的压缩方法及视频流的特点,在分析Jordan提出的基于运动矢量的视频水印算法的基础上,结合人眼视觉特点,根据视频运动矢量的幅值大小和图像纹理均匀程度,选择在帧间编码块中嵌入水印的位置,在运动矢量中嵌入水印.仿真实验结果表明,新算法能更合理地将水印嵌入Mpeg2视频流中,视频序列平均信噪比较原算法提高12%～21%,并且提取的水印要明显好于原算法.更多还原

【Abstract】 Investigates an improved motion vector based Mpeg 2 video digital watermark technique. Through analysis of the Mpeg 2 standard, combined Jordan’s algorithm which embeds digital watermark in motion vector and human eye system, watermark is located in a suitable block of inter frame according to the value of the motion vector and the character of image texture. The experimental result indicates that this method can hide the watermark in the Mpeg 2 compressed stream more rationally. The average peak SNR is improved 12%～21% than in the original algorithm, and reduces the error rate.更多还原