作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程

• 图形图像处理 • 上一篇    下一篇

基于鲁棒高阶条件随机场的视频自动分割

程婷婷,郭立君,黄元捷   

  1. (宁波大学信息科学与工程学院,浙江 宁波 315211)
  • 收稿日期:2015-01-04 出版日期:2015-07-15 发布日期:2015-07-15
  • 作者简介:程婷婷(1989-),女,硕士,主研方向:计算机视觉;郭立君,副教授、博士;黄元捷,硕士。
  • 基金资助:
    国家自然科学基金资助项目(61175026);宁波市自然科学基金资助项目(2014A610031,2014A610032);“信息与通信工程”浙江省重中之重学科开放基金资助项目(xkxl1426);宁波大学胡岚优秀博士基金资助项目(ZX2013000319);宁波大学人才工程基金资助项目(20111537)。

Automatic Video Segmentation Based on Robust Higher Order Conditional Random Field

CHENG Tingting,GUO Lijun,HUANG Yuanjie   

  1. (College of Information Science and Engineering,Ningbo University,Ningbo 315211,China)
  • Received:2015-01-04 Online:2015-07-15 Published:2015-07-15

摘要: 针对交互式分割方法存在用户标注繁琐和过分割现象,以及仅考虑二元项不能获得图像中准确的物体边界等问题,结合鲁棒高阶条件随机场,提出一种视频自动分割方法。采用基于超像素显著性特征的分割方法对视频初始帧进行自动分割,其结果作为初始化种子建立模型。根据颜色信息设计高斯混合模型,基于纹理、形状等特征,利用联合Boosting算法训练Jointboost强分类器模型,通过条件随机场提高分割准确度。引入基于超立体像素的高阶项,增加像素与区域的关联,提高分割边界的平滑度。实验结果表明,该方法明显地提高了分割效果。

关键词: 视频自动分割, 高阶势, 超立体像素, 条件随机场, 双模型融合, 特征融合

Abstract: This paper presents an automatic video segmentation method based on robust higher order Conditional Random Field(CRF),which alleviates the problem that interactive segmentation is time-consuming and labor-intensive,and oversegmentation is generated in unsupervised segmentation,and simple pairwise-pixel segmentation cannot get accurate boundary.It utilizes the saliency based segmentation of the first frame of video as initial seeds instead of user labeling.The Gaussian mixture model and a strong jointboost classifier model are respectively learned on the features of color,texture and shape,the combination of both in CRF improves the accuracy of segmentation.It adds higher order potential based on supervoxel to solve the shortcoming of oversmoothing of pairwise-pixel segmentation.Experimental results demonstrate that the method is more effective and efficient than the state-of-art methods.

Key words: automatic video segmentation, higher order potential, supervoxel, Conditional Random Field(CRF), double model fusion, feature fusion

中图分类号: