基于集群的高频合成孔径声纳并行处理方法

大测绘带下的高频合成孔径声纳需要解决的主要问题是海量数据的实时处理。该问题对数据处理的计算平台、成像处理和架构提出了很高的要求。本文利用非均匀离散快速傅里叶变换改进了成像算法,使之能够适应采用了多子阵技术的合成孔径声纳;提出了高频合成孔径声纳信号并行处理方法,在集群上实施了该方法,并进行了湖试试验。实时成像结果表明,改进的并行处理方法可以满足分辨率为距离向4cm、方位向5 cm,测绘带宽为200m的高频合成孔径声纳实时成像要求,具有较高的稳定性。

A parallel processing method for high-frequency synthetic aperture sonar based on clusters

The major issue of a Synthetic Aperture Sonar(SAS)with a large mapping width is the real-time processing of large amounts of raw data,which brings high requirement to the computing instruments,mapping algorithms and the system structure.The paper improves theω-k algorithm by using the non-uniform separate FFT to make it adapt to multi-receivers SAS. We also give a parallel method implemented on computer cluster for high-frequency SAS signal processing.This parallel processing method has been verified by a lake trial.The real-time imagining results indicate that the advanced parallel method suits for a real-time high-frequency SAS mapping need with a range resolution of 4 centimeters,an azimuth resolution of 5 centimeters and a mapping width of 200 meters,and the stability is favorable.