中文摘要：

采用超声导波评价长骨骨质状况已是近几年来的一个研究热点。在应用时频表征方法处理导波实验信号的研究中，选择交叉项抑制效果好的时频表征核函数直接影响到导波信号的分离和提取。提出将Zhao-Atlas-Marks分布（ZAMD）应用于分析长骨中的导波信号，同时与重排光滑伪维格纳维利分布（RSPWVD）的分析结果进行比较，并用图像处理中的骨架求取算法，估计导波平均频率慢度曲线。分析结果表明，ZAMD可较好地去除交叉项干扰，得到较准确的超声导波频散特性曲线，说明ZAMD是分析长骨中超声导波信号的一种较好方法。

英文摘要：

During the last years, there has been a considerable interest of using quantitative ultrasonic guided waves （GW） to evaluate the long bone quality. Time-frequency representation method is employed as a classical one for GW signal processing. Different time-frequency representations have different kernels which directly affect the separation and feature extraction. This paper employed the Zhao-Atlas-Marks distribution （ZAMD） to analyze the GW signal excited in the long bone, and comparing with reassigned smoothed-pseudo Wigner-Ville distribution （RSPWVDI. The results showed that ZAMD can remove the cross-terms effectively. Furthermore, the average GW frequency slowness curves, which are helpful for getting the quantitative dispersion characteristics of long bone, were obtained by an image processing method. In general, it is proved that ZAMD, which can smooth out the cross-terms and get the GW dispersion characteristic curves effectively, is a good one of the time-frequency methods for the GW signal processing.