中文摘要：

利用视觉注意模型提取的视频关键帧,不能全面地表达视频内容,针对这一问题,提出了基于视觉注意模型和互补性融合的关键帧提取方法。该方法利用视频帧的空间显著度和运动显著度的互补性,分别提取关键帧,再根据它们之间的相关性,消除冗余的关键帧。实验表明,该方法提取的关键帧能比较全面的表达视频内容,并且符合人类的视觉特性。

英文摘要：

As the video key frames extracted by use of visual attention model can not fully express the video content, we propose a key frame extraction method based on the visual attention model and the integration of complementary, in which the key frames are respectively extracted by means of their spatial saliency levels, and motion saliency levels, and then the redundant key frames are eliminated according to the correlation among key frames. Experiments show that the key frames completely express the video content and are consistent with the human visual properties.