声传感网中的语义增强型信息融合方法

Semantically Enhanced Information Fusion for Acoustic Sensor Networks

该文提出了一个面向声传感器网络的信息融合新方法。通过对探测到的声信号进行语义分析和自动语义属性标注,把领域知识显性化地描述出来。利用语义描述的潜在分类能力,研究了将领域专家知识引入到信息融合中两种方式。在此基础上结合传统数据融合模型,提出并构造一个将高层次语义概念引入到目标识别中的信息融合新框架。利用声传感器网络采集到的车辆声信号对所提融合方法进行了检验。仿真结果表明本方法能够在一定程度上增强车辆声识别准确性。

A new information fusionlappwach is proposed in： this paper for acoustic seffsor networks. First, by conducting semantic analysis and automatie, semantic attributes annotation for sensed acoustic signals, the do- main knowledge is described explicitly： Second, two methods to integrate domain expert knowledge into infor- mation fusion are studied, which take advantages of the potential classification capability embedded in the se- mantic description. Based on the above work and combining the traditional data fusion models, a new informa- tion fusion framework is constructed by introducing high level semantic concepts into targets recognition and tracking. In experiments, the proposed method is tested based on the vehicles＇ acoustic signals collected fio.a sensor networks. The simulation results show that the method can improve the fusion system＇ s recognition atcuracy for land vehicles＇ type.