基于聚类方法改进的关键帧提取算法

摘要

Combining with the presetting artificially of threshold and the presence of redundancy and etc.in the process of key frame extraction,an improved algorithm of key frame extraction based on clustering methods was proposed,to accurately extract key frames that can represent the main content of video shot and reduce the data processing capacity of video operation.The hierarchical clustering method was used to obtain initial clustering results.These results were optimized using the artificial immune clustering method.Corresponding number key frames were extracted.Experimental analysis results show that the algorithm can not only get more stable clustering results in the clustering process,but also extract key frames with better representation,effectively reducing the redundancy.%为准确提取代表视频镜头主要内容的关键帧,减少对视频操作的数据处理量,结合关键帧提取过程中需要人为预先设定阈值以及存在冗余等情况,提出一种基于聚类方法改进的关键帧提取算法.使用层次聚类方法获得初始聚类结果,利用人工免疫聚类方法对聚类结果进行优化,提取相应数量的关键帧.实验分析结果表明,该算法在聚类过程中得到了更稳定的聚类结果,提取出的关键帧具有很好的代表性,有效降低了冗余度.