首页> 外文期刊>Journal of information and computational science >Merging Similar Intra-shot Frames for Snapshot Based Video Retrieval
【24h】

Merging Similar Intra-shot Frames for Snapshot Based Video Retrieval

机译:合并相似的镜头内帧以进行基于快照的视频检索

获取原文
获取原文并翻译 | 示例
           

摘要

Snapshot based video retrieval is the key step of efficient camera based online-offline interactions. In this paper, we propose a similar intra-shot frames merging based solution for the snapshot based video retrieval task. Firstly, the videos are compressed into SIFT feature batches by detecting and merging similar intra-shot frames. Then the min-Hash algorithm is modified to index merged features. Finally a similarity function is designed to assess the similarity between snapshot and batch, considering the ghosting effect that usually occurs in a snapshot of video. We evaluate our proposed method in an online-offline interaction advertising system. Experimental results show that our method achieves the best, performance which verifies its effectiveness in real applications.
机译:基于快照的视频检索是基于相机的有效在线-离线交互的关键步骤。在本文中,我们为基于快照的视频检索任务提出了一种类似的基于镜头内帧合并的解决方案。首先,通过检测和合并相似的镜头内帧,将视频压缩为SIFT功能批次。然后将min-Hash算法修改为索引合并特征。最后,考虑到通常在视频快照中出现的重影效果,设计了相似度函数来评估快照和批处理之间的相似度。我们在在线-离线互动广告系统中评估我们提出的方法。实验结果表明,我们的方法达到了最佳的性能,验证了其在实际应用中的有效性。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号