一种基于改进SURF的视频DNA生成算法研究  

Study of a Video DNA Extraction Algorithm Based on Improved SURF

作　　者：徐杰[1] 何庆强[1] 许涛[1] 李乾坤[1] 田兴发[1] 孙健[1] 

机构地区：[1]电子科技大学光纤传感与通信教育部重点实验室通信与信息工程学院,四川成都611731

出　　处：《四川大学学报（工程科学版）》2016年第2期145-150,共6页Journal of Sichuan University (Engineering Science Edition)

基　　金：国家重点基础研究发展计划资助项目(2013CB329103;2012CB315905);国家自然科学基金资助项目(61201128;61271165;61301153);四川省科技支撑计划项目(2014GZ0017;16ZC1353);长江学者和创新团队发展计划项目;111引智计划项目(B14039)

摘　　要：为有效提高视频内容管理的准确性和高效性,借鉴生物信息学的方法,提出一种将降维改进的SURF算法和Kmeans算法相结合的视频DNA提取算法。该算法通过提取关键帧的SURF特征得到特征点集合,并对其进行K-means聚类构建视觉词袋模型,并将SURF特征通过视觉词袋模型量化为视觉词汇,并最终编码生成视频DNA。实验结果表明,采用改进的SURF算法生成的视频DNA具有良好的准确性和鲁棒性,并能在时间开销方面得到一定的提高。In order to improve the accuracy and efficiency of video content management,a combined method of improved SURF and Kmeans was proposed to extract the DNA of video,which referenced the bioinformatics to identify and analysis videos. By extracting SURF features on video key frames,a feature set was formed. Then,the feature set was processed by K-means,and a model of bag of words was built. Finally,SURF features were quantized visual vocabulary by the model of bag of words and encoded as video DNA. The experimental results showed that the proposed algorithm of extraction of video DNA has high accuracy and good robustness,and can make improvement in terms of time cost.

关 键 词：内容管理 特征提取 视频DNA SURF算法 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]