Debrouwere Maarten, Angland David
Faculty of Engineering and the Environment, University of Southampton, Southampton, SO16 7QF, United Kingdom.
J Acoust Soc Am. 2017 Feb;141(2):1009. doi: 10.1121/1.4976068.
Deconvolution of phased microphone array source maps is a commonly applied technique in order to improve the dynamic range and resolution of beamforming. Most deconvolution algorithms require a point spread function (PSF). In this work, it is shown that the conventional definition of the PSF, based on steering vectors, is changed when the source is rotating. The effect of rotation results in an increase in the resolution and aperture of the array. The concept of virtual array positions created by source rotation is used to derive an approximation of the PSF based on an Airy pattern. The Airy pattern approximation is suitable for use in deconvolution of rotating source maps as it is more accurate and computationally less expensive than the conventional PSF definition. The proposed Airy pattern approximation was tested with both CLEAN and DAMAS deconvolution algorithms. On the same hardware, it was significantly faster when compared to the conventional definition. The limitations of the Airy pattern approximation are shown in a synthesized broadband test case with a high dynamic range. However, in most practical beamforming applications, the proposed Airy pattern approximated PSF for deconvolution is a suitable option considering its accuracy and speed.
相控麦克风阵列源图的反卷积是一种常用技术，用于提高波束形成的动态范围和分辨率。大多数反卷积算法需要一个点扩散函数（PSF）。在这项工作中，研究表明，当源旋转时，基于导向矢量的PSF传统定义会发生变化。旋转的影响导致阵列分辨率和孔径的增加。利用源旋转产生的虚拟阵列位置概念，基于艾里斑推导出PSF的近似值。艾里斑近似适用于旋转源图的反卷积，因为它比传统的PSF定义更准确且计算成本更低。所提出的艾里斑近似在CLEAN和DAMAS反卷积算法中都进行了测试。在相同硬件上，与传统定义相比，它的速度明显更快。在一个具有高动态范围的合成宽带测试案例中展示了艾里斑近似的局限性。然而，在大多数实际波束形成应用中，考虑到其准确性和速度，所提出的用于反卷积的艾里斑近似PSF是一个合适的选择。
