• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>BioTechnology: An Indian Journal >The object segmentation technology research on tennis match video based on particle filter and a priori probability model
【24h】

The object segmentation technology research on tennis match video based on particle filter and a priori probability model

机译：基于粒子滤波和先验概率模型的网球比赛视频目标分割技术研究

获取原文
           
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Segment complete video results for a moving object can reduce the search amount of moving target and have great help for effectively enhance the tracking speed and precision. The purpose of video segmentation is to extract a moving target in video sequences from the background and to achieve segmentation of the foreground and background. The traditional method for extracting is the pixel value in the background is in accordance with the Gaussianmodel and conduct image segmentation according to 3ó rule. Although the rules can preferably extract the background, there will still be the case where foreground can be divided as the background by mistake. Therefore, this paper uses particle filter and a priori probability model to predict the moving target in the next frame, and then obtain a adaptive segmentation threshold to achieve video object segmentation according to the prediction value. This algorithm reduces the ratio that foreground points are divided into background points by mistake in the segmentation results, and it has improved for image segmentation compared with 3ó rule. Through the analysis and implementation of the algorithms, it provides a theoretical basis for the tennis tournament video object segmentation and better analyticalmethods for the technique improvement of all kinds of sports.
机译：对运动物体进行完整的视频分割，可以减少运动目标的搜索量，对有效提高跟踪速度和精度有很大帮助。视频分割的目的是从背景中提取视频序列中的运动目标，并实现前景和背景的分割。传统的提取方法是根据高斯模型提取背景中的像素值，并根据3ó规则进行图像分割。尽管规则可以优选地提取背景，但是仍然存在将前景错误地划分为背景的情况。因此，本文采用粒子滤波和先验概率模型对下一帧的运动目标进行预测，然后根据预测值获得自适应的分割阈值，实现视频对象的分割。该算法降低了分割结果中前景点被误分为背景点的比例，与3ó规则相比，在图像分割方面有所改进。通过算法的分析和实现，为网球比赛视频对象的分割提供了理论依据，为各种运动技术的改进提供了更好的分析方法。

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