互联网中对视频动态图像信息提取仿真  

Simulation of Video Dynamic Image Information Extraction in Internet

作　　者：邹航菲 徐超[1] 

机构地区：[1]江西警察学院组织人事处,江西南昌330000

出　　处：《计算机仿真》2018年第1期180-183,共4页Computer Simulation

摘　　要：对视频动态图像信息的提取问题的研究,能够有效分析互联网中视频动态图像,高效提取图像中所需信息。对互联网中的视频动态图像信息的提取,需要对相似目标镜头进行标志,选取分割阈值,获取目标视频镜头匹配值,完成图像信息提取。传统方法选择图像主分量并形成特征矢量集,并对视频镜头进行语义匹配,但忽略了获取目标视频镜头匹配值,导致提取精度偏低。提出基于视频高层语义的视频动态图像信息提取方法。利用BOVW方法提取视频目标的信息关键帧集合,计算出目标关键帧权值向量,建视频数据的语义特征库,并采用主元分析法对视频图像帧的目标高维特征进行降维,对相似目标镜头的语义进行标志,选取分割阈值,获取目标视频镜头匹配值,完成互联网中对视频动态图像信息的提取。实验结果表明,所提方法大幅度提升了视频图像信息提取信息精度。ABSTRACT：This paper proposes a method for information extraction of video dynamic image based on video high- level semantics. Firstly, the BOVW method is used to extract set of information key frame of video object and weight vector of target key frame is computed, then weight vector of semantic feature library is built and principal component analysis method is used to reduce dimensions for target high dimensional feature of video image frames. Moreover, the similar target lens is marked to select segmentation threshold and matching value of target video lens is obtained. Fi- nally, the extraction of information of video dynamic image in internet is completed. Simulation result shows that the proposed method can greatly improve the accuracy of video image information extraction.

关 键 词：互联网 视频动态图像 信息提取 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]