notesum.ai

Published at December 3

Switchable deep beamformer for high-quality and real-time passive acoustic mapping

cs.AI

cs.LG

cs.SD

eess.AS

Released Date: December 3, 2024

Authors: Yi Zeng¹, Jinwei Li¹, Hui Zhu¹, Shukuan Lu², Jianfeng Li¹, Xiran Cai¹

Aff.: ¹ShanghaiTech University; ²Xi'an Jiaotong University

Arxiv: http://arxiv.org/pdf/2412.02327v1

Refer to caption

Methods	Weight calculation
TEA [23]	$\boldsymbol{w}(\boldsymbol{r})=\frac{1}{N}\boldsymbol{1}$ (3)
RCB [9]	$\min_{\boldsymbol{a}}\boldsymbol{a}^{T}\boldsymbol{R}(\boldsymbol{r})^{-1}% \boldsymbol{a}\ s.t.\\|\boldsymbol{a}-\overline{\boldsymbol{a}}\\|^{2}\leq\varepsilon$ (4) $\boldsymbol{a}$ , $\boldsymbol{\bar{a}}$ is the actual and assumed steering vector, $\varepsilon$ is a user-defined parameter
EISRCB [10]	$\boldsymbol{w(\boldsymbol{r})}=\boldsymbol{U}_{s}\boldsymbol{U}_{s}^{T}% \boldsymbol{w(\boldsymbol{r})}_{RCB},s.t.\gamma_{1}\leq\delta\gamma_{L}$ (5) $\boldsymbol{w}_{RCB}$ is the weight of RCB $\boldsymbol{U}_{s}$ comprises the $L$ maximum eigenvalues of $\boldsymbol{R}$ , $\gamma_{1}$ , $\gamma_{L}$ are the biggest and smallest eigenvalues
DAX-RCB [12]	$\boldsymbol{w(\boldsymbol{r})}=\frac{Cov(\boldsymbol{s}_{1}(\boldsymbol{r},t),% \boldsymbol{s}_{2}(\boldsymbol{r},t))}{\sqrt{D(\boldsymbol{s}_{1}(\boldsymbol{% r},t))}\sqrt{D(\boldsymbol{s}_{2}(\boldsymbol{r},t))}}$ (6) $Cov$ , $D$ are the covariance and variance functions, respectively
RLPB [11]	$\min_{\boldsymbol{w}}\\|\boldsymbol{s}^{T}\boldsymbol{w(\boldsymbol{r})}\\|_{% \infty}\ s.t.\boldsymbol{a}^{T}\boldsymbol{w}-\tau\\|\boldsymbol{w}\\|_{\infty}\geq 1$ (7) $\tau$ is a user-defined parameter