基于压缩传感的多尺度传感器融合

Multi-scale Sensor Fusion Based on Compressed Sensing

针对传统采样理论需要采集大量观测数据所带来的存储、传输以及经济方面的压力,提出了一种基于压缩传感理论的多尺度传感器融合方法。分析了基于压缩传感理论的重构算法的设计问题,采用曲线拟合方法对多尺度的传感器的时间配准。仿真结果表明,基于正交匹配追踪算法几乎可以完全重构原始信号。与传统采样算法相比,虽然精度没有传统的方法高,但是在误差允许范围内,采集的数据少,所需时间少,减少了数据的传输和存储成本。

An algorithm of multi-scale sensor fusion based on compressed sensing theory was proposed in this paper,owing to the pressure of storage,transmission as well as economy which were caused by the traditional sampling theory.The design method of reconstruction algorithm based on the compressed sensing theory was analyzed.The curve fitting method was utilized to the time registration of multi-scale sensing fusion.The simulation results show that the algorithm of orthogonal matching pursuit could recover the original signal.To compare with the traditional sampling algorithm,the accuracy is lower.However,the cost of data transmission and storage could be reduced.