针对目前光子多普勒测速(PDV)信号处理方法抗噪性能不足的现状,提出了一种对激光多普勒信号的短时傅里叶变换(STFT)频谱图进行轨迹搜索找出有效目标信号的方法,该方法利用对搜索方向进行加权的MeanShift迭代过程来得到密度函数极大值点,然后将该点作为有效信号的轨迹点。实验表明,在给出合适的初始迭代点的情况下,该算法能准确地从含有大量噪声及干扰的短时傅里叶变换频谱图中提取出有效信号,对于强噪声条件下这种PDV信号提取算法具有较强的稳健性。 Aiming at the current situation that the anti-noise performance of photon Doppler velocimetry (PDV) signal processing method is insufficient, a tracking search of short-time Fourier transform (STFT) spectrogram of laser Doppler signal is proposed to find the effective target signal The method uses the MeanShift iterative process of weighting the search direction to obtain the maximum point of the density function and then use this point as the track point of the valid signal. Experiments show that the proposed algorithm can accurately extract the effective signal from the short-time Fourier transform spectrogram with a lot of noise and interference given the proper initial iteration points. For the PDV signal under strong noise conditions Extraction algorithm has strong robustness.