Abstract

Aiming at the problem of high complexity and poor real-time performance of the traditional minimum variance (MV) algorithm, a low-complexity minimum variance algorithm combined with power method is proposed. Firstly, the echo data is transformed into beam domain by discrete cosine transform and the dimension reduction parameter is determined according to the data of scanning lines. Secondly, the maximum eigenvalue and corresponding eigenvector of sample covariance matrix are obtained by the power method to reduce the complexity of eigenvalue decomposition. Finally, by ignoring low-energy echo signal, the inversion of covariance matrix can be simplified to construct a new weighted vector, which can reduce the complexity of MV. The Field II simulation results show that the proposed algorithm has better resolution, contrast ratio and efficiency than the traditional MV algorithm, and outperforms the minimum variance algorithm based on eigenvalue decomposition (ESBMV) in resolution and efficiency.

中文翻译：

---

低复杂度最小方差结合功率法的超声成像

摘要

针对传统最小方差（MV）算法复杂度高，实时性差的问题，提出了一种结合幂法的低复杂度最小方差算法。首先，通过离散余弦变换将回波数据变换到波束域，并根据扫描线的数据确定降维参数。其次，通过幂方法获得样本协方差矩阵的最大特征值和相应的特征向量，以降低特征值分解的复杂度。最后，通过忽略低能量回波信号，可以简化协方差矩阵的求逆，以构造新的加权矢量，从而降低MV的复杂度。Field II仿真结果表明，该算法具有较好的分辨率，