基于Snake活动轮廓模型的视频跟踪分割方法
DOI:
CSTR:
作者:
作者单位:

作者简介:

祝世平(1970-),男,博士,副教授,主要从 事图像和视频处理、计算机视觉和机器视觉在精密测量中的应用研究.

通讯作者:

中图分类号:

TP391.41

基金项目:

国家自然科学基金(61075011,60675018)和教育部留学回国人员科研启动基金资助项目 (北京航空航天大学 仪器科学与光电工程学院测控与信息技术系,北京 100191)


A video object segmentation algorithm based on Snake models
Author:
Affiliation:

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    基于Snake活动轮廓模 型,采用时空融合的方式,根据短时间内相邻帧的运动趋势差异相似 的前提,首先将视频序列分 成若干个小段,每段有k帧视频,选取段内的前两帧为关键帧, 通过运动检测的方式自动得到这两帧中运动对象的大致区域; 然后进行帧内Snake演变,搜索精确轮廓;最后以关键帧间运动对象形心的运动矢量预测 勾勒后续帧的初始轮廓,进行帧内Snake精确轮廓定位,从而实现所有帧的视频对象分割。 相比于传统方法,本文方法克服了手动绘制初始轮廓 的缺点,在空域对Snake贪婪方法进行了改进而且精确度高,速度快。实验表明,本文方法 成功地实现了前后帧图像之间运动 对象的对应匹配关系,并通过改进后的Snake贪婪方法得到了精确的分割结果。

    Abstract:

    A new video object segmentation algorithm based on the improved greedy Snake model is proposed to solve the problem of object tracking.This algorithm combines temporal and spatial information tog ether.Firstly,the video sequence can be divided into segments due to the fact that the movement trend of adjacent frames remains simi lar in a short period of time,and each segment has k frames;Secondly,the first two frames of each segment are recognized as key fra mes,and the rough contours of the moving object in the first two frames are acquired automatically by using the motion detection;Third ly,the improved inter-frame greedy Snake iteration is applied to get the precise contour;Fourthly,the intra-frame moving vectors of the moving object centers in the key frames are used to predicate the initial contours of the moving object in the subsequent frames;Fi fthly,the improved inter-frame greedy Snake iteration is applied for the non-key frames to get the precise contours on the basis of the initial contours,and then the video object segmentation can be realized for all the frames.Compared with the traditional methods,the proposed algorithm overcomes the disadvantages of drawing the initial contour manually.Furthermore,the greedy Snake method in the spatial domain has been improved with high accuracy, speed and many other obvious advantages.Experimental results indicate that the new method realizes the corresponding match of adjacent moving objects and gains accurate segmentation results through the impr oved greedy Snake method.

    参考文献
    相似文献
    引证文献
引用本文

祝世平,郭智超,高洁,马丽.基于Snake活动轮廓模型的视频跟踪分割方法[J].光电子激光,2013,(1):139~145

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2012-07-14
  • 最后修改日期:2012-09-11
  • 录用日期:
  • 在线发布日期:
  • 出版日期:
文章二维码