摘要
视频的大数据时代已经到来,将视频序列分割成镜头来进行视频内容分析和视频检索是十分重要的研究方向.文中提出一种基于帧间一致(Frame Consistency,FC)模型和光流特征的视频镜头分割技术.利用基于视觉感知的"整体到局部"的思想,首先浏览视频,除去视频的冗余信息,以降低计算成本,并通过提取视频的视觉特征构建帧间一致性函数,以此创建可能的镜头分割集合,并结合运动特征进一步优化分割结果.该技术在评估上,其精确度、召回率和F1值,都呈现出较好的效果.
Video′ s big data era has come,it is very important for research direction to divide video sequence into shots for video content analysis and video retrieval. A video shot segmentation technology based on FC (Frame Consistency) model and optical-flow feature model is proposed in this paper. The video redundant information are removed by browsing video to reduce the computational cost by means of the holistic to local thinking based on visual perception. The inter-frame consistency function is constructed by means of extracting the visual feature of video,so as to create possible lens segmentation sets and further optimize the segmentation results in comparison with the motion characteristics. The evaluation results indicate that the accuracy,recall rate and F1 value of the technology all show the good effects.
作者
高尹
刘颖
来毅
刘陆
GAO Yin;LIU Ying;LAI Yi;LIU Lu(Key Laboratory of Electronic Information Application Technology for Crime Scene Investigation,Ministry of Public Security,Xi’an 710121,China;Graduate School of Image and Information Processing,Xi’an University of Posts and Telecommunications,Xi’an 710121,China;School of Communication and Information Engineering,Xi’an University of Posts and Telecommunications,Xi’an 710121,China;International Joint Research Center of Wireless Communication and Information Processing of Shaanxi Province,Xi’an 710121,China)
出处
《现代电子技术》
北大核心
2019年第18期73-76,共4页
Modern Electronics Technique
基金
国家自然科学基金资助项目(41504115)
公安部科技强警基础工作专项项目(2016GABJC51)
陕西省教育厅资助科研项目(17JK0707)
中国科学院光谱成像技术重点实验室开放研究基金(LSIT201709D)~~
关键词
镜头分割
视觉感知
帧间一致模型
光流特征
模型评估
冗余信息去除
shot segmentation
visual perception
inter-frame consistency model
optical flow feature
model evaluation
redundant information removing
