摘要
针对增强虚拟环境(AVE)视频监控系统,提出了一种基于目标二维图像特征和三维空时特征并进行轨迹约束的运动车辆检索方法。二维检索中采用SURF特征匹配进行目标精确定位;三维检索中通过提取运动目标空时特征,充分利用AVE系统中摄像头关联信息进行关联分析,缩小目标搜索范围。实验结果表明,该算法具有较高的检索效率与精度,特别适用于多摄像头AVE监控系统中目标快速定位,掌握其在监控区域中的全空间运动信息。
For augmented virtual environment(AVE) video surveillance system,a moving target retrieval method based on trajec-tory constraints using target's 2D image feature and 3D Temporal Spatial feature is proposed in the paper.Firstly,SURF features are used for accurate positioning of target during two-dimensional retrieval process.Secondly,as for three-dimensional retrieval,the computing method of moving block's TemporalSpatial Feature for AVE surveillance system is also elaborated,then making the full use of associated information for relational analysis in order to narrow the search scope.Experimental results show that this algorithm has good performances of retrieval efficiency and accuracy,and is particularly applied to multi-camera AVE surveillance system to realize the target's quick positioning and know its full space motion information in the monitoring area.
出处
《计算机工程与设计》
CSCD
北大核心
2012年第9期3475-3480,共6页
Computer Engineering and Design
基金
中央高校基本科研专项基金项目(CCNU2009002)
武汉青年科技晨光计划基金项目(201050231066)
关键词
监控系统
视频检索
轨迹约束
运动分析
SURF特征
surveillance system
video retrieval
trajectory constraints
motion analysis
SURF