• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>NEC Journal of Advanced Technology >Integrated Object Detection and Tracking by Multiple Hypothesis Analysis
【24h】

Integrated Object Detection and Tracking by Multiple Hypothesis Analysis

机译：多重假设分析的集成目标检测与跟踪

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 this paper, we describe a novel multi-object tracking technique that integrates object detection into the object tracking process and solves the tracking problem by finding the globally optimized object trajectories through the multiple hypothesis analysis. The detection module recognizes the target objects in each frame of the video stream. The tracking module accumulates the detection results in a graph-like structure and maintains multiple hypotheses of objects trajectories. The hypotheses are ranked by their likelihoods which are computed over a sufficient number of frames, and the most likely hypothesis is used to generate the object tracking result. At the same time, the tracking module gives feedbacks to the object detection module, which are predictions of object locations in subsequent frames. Through such tight integration of the object detection and tracking, as well as the global optimization of object trajectories, we have accomplished not only robust and efficient object tracking, but also the ability to deal with occlusions, irregular object motions, changing appearances, etc. which are the challenging problems for most traditional tracking methods.
机译：在本文中，我们描述了一种新颖的多对象跟踪技术，该技术将对象检测集成到对象跟踪过程中，并通过多重假设分析找到全局优化的对象轨迹，从而解决了跟踪问题。检测模块识别视频流的每一帧中的目标对象。跟踪模块将检测结果累积为类似图形的结构，并保持对象轨迹的多个假设。假设通过在足够数量的帧上计算的可能性进行排序，最有可能的假设用于生成对象跟踪结果。同时，跟踪模块向对象检测模块提供反馈，该反馈是对后续帧中对象位置的预测。通过对象检测和跟踪的紧密结合以及对象轨迹的全局优化，我们不仅实现了强大而有效的对象跟踪，而且还具有处理遮挡，不规则对象运动，外观变化等的能力。这是大多数传统跟踪方法面临的挑战性问题。

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