基于Omega-K算法的快速全聚焦超声成像研究

Research on high-speed total focusing ultrasonic imaging method based on Omega-K algorithm

针对现有超声全聚焦技术难以实时成像的问题,提出基于w-k算法的快速全聚焦技术。首先将用于全聚焦方法（TFM）成像的三维全矩阵数据分解为N个二维子矩阵;利用快速傅里叶变换,将1～N号子矩阵由时域I（t,x）转换为频域D（kZ,kx）;基于w-k算法构建波数迁移因子F（kZ,kx）,对D（kZ,kx）进行加权得到ID（kx,kz）,实现频域中的声束聚焦;通过快速傅里叶逆变换得到子矩阵聚焦图像,并将其进行图像融合,最终获得全聚焦图像。结果表明,单核测试条件下,快速全聚焦方法获得具有200×300像素点的64阵元图像所需平均时间仅为0.65 s,而常规全聚焦算法需要1 467.36 s。综上,基于ω-k算法的全聚焦技术具有成像速度快、对硬件要求低等优点,为实时的高精度在线无损检测提供了一种可行方法。

Aiming at the problem that existing ultrasonic total focusing imaging technique is difficult to achieve real time imaging, this paper presents a high-speed ultrasonic total focusing method（TFM） based on ω-k algorithm. Firstly, the three-dimensional full matrix data used for total focusing imaging are decomposed into N two-dimensional sub-matrices. The number 1-N sub-matrix is transformed from time domain I（t, x） into frequency domain D（kZ, kx） with fast Fourier transform（FFT）. The wavenumber migration factors are established using ω-k algorithm and are used to perform weighting to the D（kZ, kx） and obtain ID（kx, kz）. Then the sound beam focusing in frequency domain is realized. The focus images for the sub-matrices are obtained using the inverse FFT（IFFT）. At last, the TFM image is established by fusing all the focus images. The result shows that under the test condition of single CPU core, the average time for the TFM based on ω-k algorithm to obtain a 64 array element image with 200×300 pixels is only 0.65 s. However, the conventional TFM consumes 1 467.36 s in the same condition. In summary, the TFM based on the ω-k algorithm has the advantages of high-speed operation and low hardware requirement, which provides a feasible method for the on line real-time high-precision non-destructive testing.