基于子镜头分割的自适应关键帧提取算法.pdfVIP

中国科技论文在线 基于子镜头分割的自适应关键帧提取算法 段豪，谢刚，雷少帅，杨倩** （太原理工大学信息工程学院，太原 030024） 5 摘要：关键帧的提取是基于内容的视频检索的重要技术，它也是视频分析和基于高级语义的 视频检索的基础。本文提出了一种基于子镜头分割的自适应关键帧提取算法。该方法首先利 用样本分类中的距离可分性准则构建分割判别函数，随后通过探讨判别函数的极值来实现子 镜头分割，最后根据每个子镜头内容的变化自适应地提取相应数目的关键帧。实验结果表明， 该算法提取的关键帧不仅保持了视频的时序性，而且能以较低的冗余度实现对视频内容的准 10 确描述。 关键词：关键帧提取；子镜头；距离可分性；自适应 中图分类号：TP391 Sub-shot Segmentation-Based Self-Adaptive Key Frame 15 Extraction Algorithm DUAN Hao, XIE Gang, LEI Shaoshuai, YANG Qian (College of Information Engineering,Taiyuan University of Technology, TaiYuan 030024) Abstract: Key frame extraction is an important technology for content-based video retrieval, which is also the foundation of the video analysis and the video retrieval based on advanced 20 semantic. A self-adaptive key frame extraction algorithm based on sub-shot segmentation is proposed in this paper. The segmentation discriminate function utilizing distance separability criteria of sample classification is constructed firstly, and then sub-shot segmentation is achieved by exploring the local maximum of discriminant function. Finally, the key frame is adaptive selected by analyzing the sub-shot content change. Experimental results show that, the key frame 25 extracted by the algorithm not only to maintain the video frame timing, but also can achieve accurate description of video content with lower redundancy. Keywords: Key frame extraction; Sub-shots; Distance separa