...
  • 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>Signal processing >STV-based video feature processing for action recognition
【24h】

STV-based video feature processing for action recognition

机译：基于STV的视频特征处理以进行动作识别

获取原文
获取原文并翻译 | 示例
           
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

In comparison to still image-based processes, video features can provide rich and intuitive information about dynamic events occurred over a period of time, such as human actions, crowd behaviours, and other subject pattern changes. Although substantial progresses have been made in the last decade on image processing and seen its successful applications in face matching and object recognition, video-based event detection still remains one of the most difficult challenges in computer vision research due to its complex continuous or discrete input signals, arbitrary dynamic feature definitions, and the often ambiguous analytical methods. In this paper, a Spatio-Temporal Volume (STV) and region intersection (RI) based 3D shape-matching method has been proposed to facilitate the definition and recognition of human actions recorded in videos. The distinctive characteristics and the performance gain of the devised approach stemmed from a coefficient factor-boosted 3D region intersection and matching mechanism developed in this research. This paper also reported the investigation into techniques for efficient STV data filtering to reduce the amount of voxels (volumetric-pixels) that need to be processed in each operational cycle in the implemented system. The encouraging features and improvements on the operational performance registered in the experiments have been discussed at the end.
机译：与基于静止图像的过程相比，视频功能可以提供有关一段时间内发生的动态事件的丰富而直观的信息，例如人类行为，人群行为以及其他主题模式的变化。尽管在过去的十年中图像处理取得了长足的进步，并看到了其在面部匹配和目标识别中的成功应用，但是基于视频的事件检测仍然是计算机视觉研究中最困难的挑战之一，因为其复杂的连续或离散输入信号，任意动态特征定义以及通常不明确的分析方法。在本文中，提出了一种基于时空体积（STV）和区域相交（RI）的3D形状匹配方法，以帮助定义和识别视频中记录的人类动作。该方法的显着特点和性能增益源于本研究开发的系数因子增强的3D区域相交和匹配机制。本文还报告了对有效STV数据过滤技术的研究，以减少所实现系统中每个操作周期中需要处理的体素（体积像素）的数量。最后讨论了在实验中记录的令人鼓舞的功能和对操作性能的改进。

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