基于K均值聚类的视频关键帧提取技术研究  

Research on video Key Frame Extraction Based on K-means Clustering

作　　者：张一凡 李家辰 旷远有 刘盼 赵子元 ZHANG Yi-fan;LI Jia-chen;KUANG Yuan-you;LIU Pan;ZHAO Zi-yuan(School of Information Science and Technology,North China University of Technology,Beijing 100144,China)

机构地区：[1]北方工业大学信息学院,北京100144

出　　处：《电脑与信息技术》2021年第1期13-16,共4页Computer and Information Technology

基　　金：北京市大学生科学研究与创业行动计划资助。

摘　　要：文章提出了一种基于k均值聚类的视频关键帧提取算法。该算法在视频的镜头分割算法之上,通过层次聚类对视频内容进行初步划分,之后结合k均值聚类算法对初步提取的关键帧进行直方图特征对比去掉冗余帧,最终确定视频的关键帧序列。经过大量的实验数据证明,该算法能够提取出冗余度较低、代表性较强的关键帧,一定程度上为视频关键帧提取提供了有效的方法。Based on the k-means clustering method,this paper proposes an algorithm for video key frame extraction.The algorithm,including video shots segmentation algorithm,partitions the video content through hierarchical clustering,then combined with k-means clustering algorithm for preliminary extract key frames histogram feature comparison to remove redundant frames,ultimately determine the key frames of video sequences.After a lot of testing,it is proved that the key frame redundancy extracted by this algorithm is relatively small,with high representativeness,and has certain validity of video key frame extraction.

关 键 词：视频镜头分割 关键帧提取 K均值聚类 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]