利用阈值多谱图去除Wigner分布的交叉项

Removing cross-terms from Wigner distribution via thresholding multiple spectrograms

提出了一种阈值多谱图估计WD自项支撑区域,从而去除WD交叉项的新方法。所获得的TF分布拥有好的TF聚集性和极少的交叉项。WD自项支撑是从具有高斯线性调频窗(GLCW)的谱图集合中估计出来的。通过形态学滤波估计出所有谱图的自项支撑,然后所有自项支撑的交集用来作为WD的自项支撑估计。此外对于具有GLCW的线性调频信号的谱图集合的联合分辨力进行了详细的讨论。实验结果说明了该方法的良好性能。

A new method is proposed to remove cross-terms from the WD by thresholding multiple spectrograms, which is based on estimation of auto-term support of the WD. The obtained TF distribution has high TF concentration and few cross-terms. The auto-term support of the WD is estimated from a set of spectrograms with Gaussian linear chirp windows （GLCW＇ s）, where the auto-term supports of all the spectrograms are estimated by thresholding the spectrograms followed by morphological filtering, and then the intersection set of all the estimated supports is used as the auto-term support estimate of the WD. On the other hand, the joint resolution power of a set of spectrograms with GLCW＇ s for linear frequency modulation signals is discussed and defined in detail. Some experimental results are provided to demonstrate the good performance of the method.