摘要
为了改进现有关键帧提取方法在相邻帧间相似性测度方面的不足,提出一种基于MPEG-7颜色布局特征匹配和块运动信息度量的关键帧选取方法。给出颜色布局特征的提取、相似度计算方法以及利用图像分块计算运动信息的方法,并结合距离累加算法提取视频镜头关键帧。实验结果表明,本方法比基于图像综合特征关键帧提取方法检索性能有明显提高,不易产生冗余和漏检,提取的关键帧能够充分代表镜头的内容。
In order to improve the deficiency of comparability measure between the frames in the recent method of key-frame extracting, a new method of key-frame extracting based on color layout feature and block-moving information of MPEG-7 is proposed in this paper. This paper gives the method of extracting the color layout feature and matching them, and gives the method of computing the block-moving information by using image segmentation scheme, and then, it extracts the key-frames by combing the cumulative distance of detection methods. The experiment results show that our method is more accurate than the method, which uses the integrated feature of image to extract the key-frames, and doesn't easy to have redundancy and undetected, and the extracted key-frames make a good representation for video contents.
出处
《广西大学学报(自然科学版)》
CAS
CSCD
北大核心
2010年第2期310-314,共5页
Journal of Guangxi University(Natural Science Edition)
基金
广西科学基金资助项目(桂科自0991058)
广西高校人才小高地建设创新团队资助计划项目(桂教人[2007]71号)