中文摘要：

为了视频版权保护的需要,设计了一种基于改进人工蜂群算法的视频水印方案.针对传统人工蜂群算法中收敛速度慢和容易陷入局部最优的缺点,通过自适应的位置更新、信息素与灵敏度配合选择模型以及最差解替代等方法改进算法.通过利用改进后的人工蜂群算法实现一种基于独立分量分析的运动目标检测,提取出视频序列中的运动分量.对检测到的运动目标进行2阶小波变换来嵌入水印,运动目标是视频序列中固有的稳定的重要特征,将水印嵌入运动目标中提高了鲁棒性.仿真结果表明,采用检测出清晰的运动轨迹嵌入水印信息,达到了更好的不可见性;针对多种视频攻击也有很好的鲁棒性.

英文摘要：

To meet the need of video copyright protection,a video watermarking scheme based on improved artificial bee colony algorithm was designed.First,aiming at overcoming the shortages of traditional artificial bee colony algorithm including slow convergence and falling into the local optimum easily,adaptive location update,selection model with the pheromone and the sensitivity as well as the worst solution replacing were introduced into the improvement of the algorithm.Second,motion objects were detected by motion detection method based on independent component analysis of improved artificial bee colony algorithm,the motion component were extracted from video sequences.Last,watermark was embedded in motion objects which were transformed by wavelet transform.The result indicated that the watermark embedded in clear moving objects trajectory was more invisible and had strong robustness for a variety of video attacks.