摘要: 提出了一种联合时间域和空间域的视频对象提取算法。在无须事先对视频对象的时-空域信息具有先验知识的情况下,采用半自动算法,只需要有限的用户交互,通过变化检测、帧内分割和对象跟踪3个模块,对视频对象进行检测和跟踪。试验表明,尽管所跟踪的对象经历了各种变化,仍能得到一个很好的结果。
关键词:
视频对象提取,
图像分割,
运动矢量估计,
分水岭算法,
变化检测
Abstract: This paper presents a video object extraction algorithm corcerning both the temporal and spatial information. With limited human interface, the extraction system can efficiently extract the object out of the video sequence through changing detection, inter-frame segmentation and object tracking. Experiment shows that the system has achieved a qualified tracking result.
中图分类号:
梁天明,汤迪逊,陈健. 视频对象提取系统的研究与实现[J]. , 2003, 29(01): 191-.
LIANG Tianming, TANG Dixun,CHEN Jian. Research and Implementation of a Video Object Extraction System[J]. , 2003, 29(01): 191-.