3D视频目标分割与快速跟踪
Object segmentation and fast tracking in 3D video
朱仲杰 1张慧敏 1王玉儿2
作者信息
- 1. 浙江万里学院宁波市DSP重点实验室,浙江宁波315100;郑州大学物理工程学院,河南郑州450001
- 2. 浙江万里学院宁波市DSP重点实验室,浙江宁波315100
- 折叠
摘要
提出一种面向3D视频的目标分割与快速跟踪算法,主要分为空域分割和时域与视域跟踪两个主要步骤.空域分割采用基于改进C-V模型的方法.首先基于帧差法提取运动区域作为目标的初始轮廓,以减少迭代次数加快收敛速度.然后在传统的C-V模型中加入图像边缘信息以加强C-V模型对模糊边缘的处理能力,提高分割结果的主管视觉效果.时域和视域的目标跟踪采用基于改进生物地理优化的快速跟踪方法.实验结果表明所提分割与跟踪算法快速、高效.
Abstract
A new algorithm for object segmentation and fast tracking in 3D video is proposed which mainly consists of two steps:spatial segmentation and object tracking along temporal and view directions.Spatial segmentation is implemented based on an improved C-V model.Firstly,moving regions are detected based on frame difference and the results are used as initial contours of C-V model aiming to reduce the iteration times and accelerate the convergence rate.Secondly,an energy term is added into the original C-V model to enhance its ability to process blur edges and therefore to improve the subjective visual effect of the overall segmentation results.The object tracking along temporal and view directions is performed based on an improved biogeography-based optimization method.Experiments are performed and convincing results are obtained.
关键词
C-V模型/3D视频/目标分割/快速跟踪Key words
C-V model/3D video/object segmentation/fast tracking引用本文复制引用
基金项目
国家自然科学基金(60902066)
国家自然科学基金(60872094)
国家自然科学基金(60832003)
浙江省自然科学基金(Y107740)
浙江省教育厅科学研究项目(Y201016875)
宁波市自然科学基金(2012A610040)
出版年
2013
NETL
NSTL
维普
万方数据