多子阵合成孔径声呐波数域算法不均匀采样问题研究

Research on non-uniform sampling problem when adapting wave number algorithm to multiple-receiver synthetic aperture sonar

多子阵合成孔径声呐(SAS)方位向采样不均匀,因而单子阵波数域算法(ω-k)不能直接用于多子阵SAS图像重建。为此,提出两种改进ω-k算法的方法,使其可以应用于多子阵SAS图像重建。不均匀分离快速傅里叶变换(NSFFT)方法采用多子阵匀速直线运动的假设,将方位向不均匀采样的傅里叶变换分解为若干个均匀采样的傅里叶变换;不均匀快速傅里叶变换(NFFT)方法则直接快速计算方位向不均匀采样的傅里叶变换。文中对这两种方法进行了理论分析,并利用仿真数据和湖试数据对这两种方法进行了验证。成像结果表明两种方法均可用于多子阵SAS系统图像重建。NSFFT计算效率较高,但声呐基阵前进方向速度的不均匀性会对成像质量造成一定的影响;NFFT可以适用于任意速度的情况,但计算效率比NSFFT方法低。综合来看,两种方法各有特点,可以根据应用场景的不同进行选择。

The azimuth sampling of multiple-receiver SAS （Synthetic Aperture Sonar） system is non-uniform, which causes standard wavenumber algorithm （ω - k） can＇t be applied to multiple-receiver SAS image reconstruction. To solve the problem, two methods are presented, which can adapt the standard ω - k algorithm to multiple-receiver SAS system. One method named Non-uniform Separate Fourier Transform （NSFFT） converts the Fourier Transform （FT） of the non-uniform sampling in azimuth direction into several uniform FTs on the assumption that the sonar array moves along a linear track in a uniform velocity. The other method, however, calculates the FT of non-uniform sampling by non-uniform fast FT （NFFT）. Detail analysis is presented on two methods＇ theoretical principles. For validation, both methods are applied to reconstruct images of the simulation datasets and the lake-trial datasets. Results show that both methods can be applied to image reconstruction of multiple-receiver SAS. The NSFFT method has the advantage of higher computing efficiency but produces lower quality images when velocities of the sonar array vary in a large range. By contrast, the NFFT method is able to deal with arbitrary-velocity variation but has a heavier computing load. In conclusion, both methods have pros and cons, and the choice of two methods is determined by the application situation.