A High-Resolution Minimum Variance Algorithm Based On Optimal Frequency-Domain Segmentation

In order to enhance the adaptability of minimum variance (MV) algorithm to broadband ultrasound signals, a high-resolution MV beamforming algorithm based on optimal frequency-domain segmentation is proposed. By an improved time-frequency concentration criterion based on logarithmic window energy, the optimal window length of Short-time Fourier Transform (STFTMV) is obtained, and the ultrasound signals are converted into narrowband sub-signals through optimal frequency-domain segmentation. Then, the sub-signals in frequency domain are reconstructed according to the non-overlapping characteristic of window, which further improves the imaging quality and extends the limitations of traditional broadband MV near-field imaging. Moreover, according to the symmetry of STFTMV, the computation is reduced by half, which further improves the efficiency. The Field II results indicate that the mainlobe width of the STFTMV is reduced to 38.09 % of MV and 47.05 % of eigenspace-based minimum variance (ESBMV), and the imaging efficiency of STFTMV is almost 2.5 times higher than that of ESBMV. In addition, the contrast of the proposed STFTMV can be further improved by combining with the improved coherence factor (FCF) based on sub-band frequency signals.