摘要
提取视频序列中的目标轮廓,进行定位、跟踪,是当前视频处理技术研究中的一个热点。提出了一种新的视频目标半自动定位跟踪的算法。对于初始帧,采用改进的交互方法进行目标的精确定位;在后续帧中,采用零交叉点约束的Snake算法,对具有凹边和尖点特征的复杂目标准确跟踪,获得了理想的实验结果。
It is a hot topic how to extract video objects from a large amount of video data in both spatial and temporal domains. In this paper, a semi-automatic algorithm to locate an object in the video is proposed. For the first frame of the video, the object edge is obtained by an enhanced interactive method; in the remaining frames of the video, a new Snake algorithm is proposed to detect the sharp-angled and concave object contour under the constraint of zero-crossing point, the experimental results of different video objects are obtained.
出处
《兵工学报》
EI
CAS
CSCD
北大核心
2005年第5期615-619,共5页
Acta Armamentarii
基金
国家自然科学基金项目(60473141)
"十五"国防预先研究项目(41101050403)
