基于结构相似度聚类视频关键帧提取方法  

Extraction Method of Video Key Frame Based on Structural Similarity Clustering

作　　者：魏子怡 任小玲[1] 孟玉茹 WEI Ziyi;REN Xiaoling;MENG Yuru(School of Computer Science,Xi'an Polytechnic University,Xi'an 710048)

机构地区：[1]西安工程大学计算机科学学院,西安710048

出　　处：《计算机与数字工程》2024年第10期3153-3157,共5页Computer & Digital Engineering

摘　　要：针对帧间差分法对视频关键帧提取冗余度高的问题,提出了基于结构相似度聚类视频关键帧提取方法。首先,通过帧间差分法对视频内容进行粗划分;其次利用层次聚类法对粗划分的结果进行聚类;最后,引用结构相似度函数(SSIM)剔除其冗余帧,得到最终的关键帧序列。实验结果表明,该方法与传统的帧间差分法及颜色直方图方法相比,能够更好地提取出关键帧,且平均查全率和查准率分别提了10.23%、5.33%。In order to solve the problem of high redundancy in the extraction of video key frames by inter-frame difference method,an extraction method of video key frames based on structural similarity clustering is proposed.Firstly,the video content is roughly divided by the inter-frame difference method.Secondly,the result of the coarse division is used for clustering,and the structural similarity function(SSIM)is introduced to eliminate redundant frames to obtain the final key frame sequence.The experi-mental results show that compared with the traditional inter-frame difference method and color histogram method,the method in this paper can better extract key frames,and the average recall and precision rates are improved by 10.23%and 5.33%,respective-ly.

关 键 词：关键帧 帧差法 聚类 结构相似度 

分 类 号：TP393.4[自动化与计算机技术—计算机应用技术]