基于MPI并行计算的卫星遥感数据变化监测系统设计

Design of Satellite Remote Sensing Data Change MonitoringSystem Based on MPI Parallel Computing

为控制卫星遥感数据变化指征与非变化指征之间的实值误差,实现对卫星遥感数据传输变化行为的准确监测,设计基于信息传递接口(message passing interface,MPI)并行计算的卫星遥感数据变化监测系统;建立模拟/数字(analog/digital,A/D)转换电路与Zigbee网络节点的匹配关系,将所得卫星遥感数据按需分配给传感器模块与协调器设备,完成数据采集与监测终端的硬件设计;按照并行拓扑结构连接形式,实施MPI的基本调用,再联合数据进程编写原则,实现对MPI并行计算的完善;根据传感器几何校正原理,求取遥感数据度量值,再分析核函数表达式,完成对卫星遥感数据的处理;利用MPI并行计算,控制数据采集与监测终端的硬件执行设备,完成卫星遥感数据变化监测系统设计;实验结果表明,所设计系统的卫星遥感数据变化指征与非变化指征之间的实值误差的均值为0.36像素/s,与对照组相比,明显缩小了2.42像素/s,能够准确监测卫星遥感数据传输变化行为。

In order to control the real-value error between the change indicators and non-change indicators of satellite remote sensing data, and to achieve accurate monitoring of the transmission and change behavior of satellite remote sensing data, a satellite remote sensing data change monitoring system based on message passing interface(MPI) parallel computing is designed. Establish the matching relationship between the analog/digital(A/D) conversion circuit and the Zigbee network node, distribute the obtained satellite remote sensing data to the sensor module and the coordinator device as needed, and complete the hardware design of the data acquisition and monitoring terminal. According to the connection form of the parallel topology, the basic call of MPI is implemented, and the principle of data process writing is combined to realize the improvement of the parallel computing of MPI. According to the principle of sensor geometric correction, the metric value of remote sensing data is obtained, and then the kernel function expression is analyzed to complete the processing of satellite remote sensing data. Using MPI parallel computing to control the hardware execution equipment of the data acquisition and monitoring terminal, the design of the satellite remote sensing data change monitoring system is completed. The experimental results show that, the average value of the real-valued error between the change indicators and non-change indicators of the satellite remote sensing data of the designed system is 0.36 pixels/s, compared with the control group, it is significantly reduced by 2.42 pixels/s, which can accurately monitor the change behavior of satellite remote sensing data transmission.