云计算下基于贝叶斯网络的多传感器目标识别

Multi-sensor target recognition based on Bayesian classifier in cloud computing environment

摘要:

基于云计算可实现分布式并行程序海量数据处理的特点，提出将多传感器目标识别融合处理部署在云计算Hadoop平台上，并将其运行在多个节点组成的计算机集群上。根据目标识别原理建立贝叶斯网络结构，对目标识别预处理得到的数据进行融合计算，推理目标类型，并对不同情况下的Hadoop集群效率进行分析比对。实验结果证明了将目标识别融合处理部署在云计算平台上可有效提升运算效率。

Abstract:

Based on the cloud computing characteristics of distributed parallel processing of massive data，multi-sensor target recognition on the Hadoop platform is put forward, which consists of a plurality of nodes running on computer clusters. Bayesian network structure is established according to target recognition theory. The data obtained by the pre-process of object recognition are fusion-calculated to deduce the target type. The efficiencies of Hadoop clusters under different circumstances are compared. Experimental results show that running multi-sensor information fusion target recognition on cloud computing platform is intuitive and effective, which successfully improves the operational efficiency.