利用自适应阈值的聚类算法实现关键帧提取

Realization of Key Frame Distilling Based on Cluster Algorithm of Adaptive Threshold

聚类算法是关键帧提取的一种常用方法,然而,在对视频内容一无所知的情况下,预先指定聚类算法的阈值是一个很困难的问题,文章首先提出了一种自适应确定聚类阈值的算法,再以视频中的帧序列为数据样本进行聚类,得到最终聚类结果,最后从每一类中提取离类中心最近的帧作为关键帧。实验表明,该方法能较好的提取出视频序列的关键帧。

The cluster algorithm is the common method to distill the key frame. But it is problematic to predefine the absolute number of key frames without knowing the video content. In the paper, a method to confirm the adaptive threshold is first put forward, then the frame orders in the video are listed as data sample and clustered, the final cluster algorithm is obtained, then the frame nearest to the center of its class is automatically distilled as one key frame in each class. Experimental results demonstrate that the method can better distill the key frame in the video order.