基于M-Chord的分布式视觉监控交互检索算法  

M-Chord based distributed interactive visual surveillance retrieval algorithm

作　　者：周凡[1] 曲琳[1] 陈耀武[1] 

机构地区：[1]浙江大学数字技术及仪器研究所,杭州310027

出　　处：《哈尔滨工业大学学报》2009年第5期125-129,共5页Journal of Harbin Institute of Technology

基　　金：浙江省科技计划重大科技攻关资助项目(2005C11001-02)

摘　　要：为对城域视觉监控系统产生的海量多媒体数据进行高效索引和检索,提出基于M-Chord的分布式交互检索和负载均衡算法.对象以颜色柱状图、Gabor纹理描述和轨迹系数空间特征组成的特征向量描述,由M-Chord算法进行分布式索引分发和检索.提出基于线性和非线性支持向量机的分布式相关反馈算法,将支持向量机模型建立的度量空间与M-Chord索引所处的度量空间结合,利用分片中心点减少了对节点和对象的访问.实验表明提出的算法可以准确反映用户的查询语义,仅需要检索分布式监控网络中的少部分节点即可达到与顺序扫描接近的查准率,同时较好实现了分布式监控网络中的负载均衡.An M-Chord based distributed interactive retrieval and load balance algorithm is proposed to effectively index and retrieve the mass multimedia data produced by metropolitan area visual surveillance system. The object was represented by a feature vector which includes color histogram, Gabor texture descriptor and coefficient space trajectory descriptor. The feature vector was then distributed to the DHT network by the M- Chord algorithm. To distribute and retrieve the objects, distributed relevance feedback algorithms based on both linear and non-linear SVM were proposed. The metric space constructed by the SVM model was treated as the M-Chord indexing data space and the number of accesses to the nodes and objects was decreased by the pivots in the indexing data space. Experimental results show that the algorithm proposed can exactly match the semantic query of users and achieve a precision comparable to the sequential scan algorithm by checking only few nodes in the distributed surveillance network. Furthermore, the balance in the distributed surveillance network.

关 键 词：视觉监控 相关反馈 支持向量机 分布式检索 负载均衡 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]