基于互信息熵和NetVLAD的视频关键帧提取方法
Video Keyframe Extraction Based on Mutual Information Entropy and NetVLAD
康佳慧 1纪松 2范大昭 2储光涵 2李林林3
作者信息
- 1. 信息工程大学地理空间信息学院,河南 郑州,450001;嵩山实验室,河南 郑州,450046
- 2. 信息工程大学地理空间信息学院,河南 郑州,450001
- 3. 河南测绘职业学院,河南 郑州,450015
- 折叠
摘要
针对现有关键帧提取方法时间复杂度高、漏检率大、忽略视频语义信息等问题,提出一种基于互信息熵和局部聚合描述符向量网络(vector of local aggregated descriptors net,NetVLAD)的视频关键帧提取方法.首先计算视频帧互信息熵,将视频划分为视频子集;然后通过NetVLAD进行视频帧的特征提取与聚类,根据最近邻匹配算法计算帧间距离,提取候选关键帧;最后通过感知哈希减少冗余度,得到关键帧集合.基于UAV-123数据集进行了实验分析,结果表明,该方法高鲁棒地提高了关键帧的提取效率,保证了高保真度的同时降低了关键帧的冗余.
Abstract
To solve the problems of existing key frame extrac-tion methods,such as high time complexity,high miss rate and video semantic information neglect,we propose a video keyframe extraction method based on mutual information en-tropy and vector of local aggregated descriptors net(NetV-LAD).First,we calculate the mutual information entropy of video frames and divide the video into video subsets.Then,feature extraction and clustering of video frames are carried out by NetVLAD.The similarity between frames is calcu-lated by the nearest neighbor matching algorithm,and candi-date keyframes are extracted.Finally,the redundancy is re-duced by perceptual hashing,and the keyframe set is ob-tained.Experimental analysis based on UAV-123 data set proves that the proposed method improves the extraction effi-ciency of keyframes with high robustness and reduces the re-dundancy of key frames with high fidelity.
关键词
视频关键帧/互信息熵/局部聚合描述符/感知哈希Key words
video keyframe/mutual information entropy/lo-cal aggregated descriptor/perceptual hashing引用本文复制引用
基金项目
国家自然科学基金(41971427)
&&(42-Y30B04-9001-19/21)
河南省科技重大专项管理体系嵩山实验室项目(221100211000-4)
出版年
2024
NETL
NSTL
万方数据