基于GPU的线频调变标算法并行实现

Fast Implementation of Chirp Scaling Imaging Algorithm Based on GPU

论文提出了一种基于图形处理器的并行线频调变标成像方法,为合成孔径声纳实时成像提供了新的手段。通过对合成孔径声纳线频调变标成像算法中的回波转化、方位向和距离向FFT变换以及点乘运算的并行化设计,实现了图形处理器环境下的多子阵合成孔径声纳线频调变标成像算法。对真实合成孔径声纳原始回波的成像试验结果表明:并行成像算法在保持高精度成像的同时加速比高达56.25,满足合成孔径声纳系统实时成像需要。

A parallel chirp scaling(CS)imaging algorithm based on graphic processing unit(GPU)is proposed for multi-receiver synthetic aperture sonar(SAS),which provides a new method of real-time imaging for SAS. The echo transformation,azimuth and range FFT transformation,and point multiplication of CS algorithm for SAS are designed parallel in GPU environment. The imaging results for the real echo of synthetic aperture sonar show that the parallel CS algorithm keeps the high precision of SAS imaging result and the speedup reaches 56.25,which makes it meet the demand of SAS system for real-time imaging.