HBase中基于时空特征的监测视频大数据关联查询研究

Research on associated query of monitoring video big data based on spatio-temporal characteristics in HBase

针对传统的时空索引构建、维护困难且实时查询效率低等问题,提出基于HBase的时空索引构造方法。该方法采用HBase作为监测视频大数据时空特征索引结构,通过Z填充曲线对空间特征进行降维存储,并利用时间、空间与属性特征之间的关联及依赖规则来安排rowkey索引键,可有效解决传统的时空索引构建、维护困难的缺陷。针对传统的时空索引实时查询效率低的问题,提出了基于Z曲线的时空关联查询算法。该算法对查询空间计算Z值范围和建立空间划分子集,利用划分后的时空特征进行列索引查询得到候选数据集并反查HBase索引表完成关联查询。实验结果表明,与传统的R树索引算法相比,提出的基于HBase的时空索引构造方法索引插入效率更高,提出的基于Z曲线的时空关联查询算法能够快速高效地处理时空关联查询。

Aiming at solving problems like the difficulties in bui lding and maintaining of the traditionaand the inef ficiency of real-time query, this paper first proposed a new building method of spatio-temporal index based on HBase. By using HBase as the spatio-temporal features index structure of monitoring video big data,using Z-filling curve tofulfill the dimension reduction and storage of spatio-temporal features,and using the associated relations and rule between time,space and at tributive characters to arrange the rowkey index key,these disadvantages like the traditional spatio-temporal index5 s bui lding and maintaining would be overcome eff iciently. In addi ttraditional space-time index low query efficiency issues in real time,this paper further proposed spatio-temporal query algorithm based on the Z curve , which could calculate Z value ranges and establish space to subset partiocandidate data sets would be obtained by using the spatio-temporal features got from the first step to haand next the related query would be realized by pegging the HBase index table. Experimental compared with the conventional R-tree indexing algorithm, the spatio-temporal index building method based on posed above has higher insertion index efficiency and the association Z curve space-time query-based algorithm propoq uickly and efficiently deal with spatio-temporal correlated queries.