首页> 外国专利> An Effective Object Tracking Method and for Apparatus for Interactive HyperLink Video

An Effective Object Tracking Method and for Apparatus for Interactive HyperLink Video

机译:一种有效的交互式超链接视频对象跟踪方法和装置

摘要

PURPOSE: An apparatus and a method for effectively tracking random objects for a hyper-link are provided to maximize the effect of a manufacturing of a moving picture to be used in the hyper-link video by effectively tracking objects irrespective of changes of a rotation and a size of objects and outputting tracked information and various information for displaying locations of objects. CONSTITUTION: An object to be tracked is selected, or a location is extracted in a current frame of a previously inputted object(S402). A characteristic point of the object to be tracked is extracted(S403). A noise is removed using a median filter for removing a noise of a video, and an edge is extracted using the Canny algorithm(S403). A frame data obtained in the first buffer and the second buffer is analyzed, and a motion vector of characteristic points of the obtained object is extracted(S404). A kind of light having a same direction is classified using the motion vector of each characteristic point(S405). The classification having the most kind of light is selected(S406). A direction having the most kind of light is selected among the selected direction and a moving direction of the object is determined for determining a location of the object(S407). A mean of a motion vector is determined as a moving location of the object(S408). It is judged whether a current frame is an end frame of a frame in which a scene is changed before a location of the object is tracked in a next frame(S409). If the current frame is not the end frame, it is judged whether the change of the rotation or the size of the object is occurred in the current determined location of the object(S410).
机译:目的:提供一种用于有效跟踪超链接的随机对象的装置和方法,以通过有效地跟踪对象而与旋转和旋转的变化无关地最大化在超链接视频中使用的运动图像的制造效果。物体的尺寸并输出跟踪的信息和用于显示物体位置的各种信息。构成:选择要跟踪的对象,或在先前输入的对象的当前帧中提取位置(S402)。提取要跟踪的对象的特征点(S403)。使用用于去除视频的噪声的中值滤波器去除噪声,并且使用Canny算法提取边缘(S403)。分析在第一缓冲器和第二缓冲器中获得的帧数据,并且提取获得的对象的特征点的运动矢量(S404)。使用每个特征点的运动矢量对具有相同方向的光进行分类(S405)。选择具有最多种类的光的分类(S406)。在选择的方向中选择具有最多种类的光的方向,并确定对象的移动方向以确定对象的位置(S407)。将运动矢量的平均值确定为对象的移动位置(S408)。判断当前帧是否是在下一帧中跟踪对象的位置之前场景发生改变的帧的结束帧(S409)。如果当前帧不是结束帧,则判断在对象的当前确定位置中是否发生了对象的旋转或尺寸改变(S410)。

著录项

  • 公开/公告号KR20010035100A

    专利类型

  • 公开/公告日2001-05-07

    原文格式PDF

  • 申请/专利权人 VIRTUALMEDIA;

    申请/专利号KR20000080967

  • 发明设计人 LEE SEONG HWAN;SON HYEONG JIN;

    申请日2000-12-22

  • 分类号H04N7/32;

  • 国家 KR

  • 入库时间 2022-08-22 01:13:46

相似文献

  • 专利
  • 外文文献
  • 中文文献
获取专利

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