中文摘要：

与短时傅里叶变换、连续小波变换、广义S变换等时频分析方法相比，匹配追踪方法具有更高的时频分辨率，但传统的贪婪迭代算法计算效率较低。以Morlet小波作为时频原子进行匹配追踪，通过分析尺度因子不同时Morlet小波时频原子在时间域的形态，比较信号向频率、相位和延时相同，仅尺度因子不同的不同时频原子投影的投影值，认为尺度因子对时频原子的形态具有较强的控制作用，因而对时频原子和信号局部特征的匹配性能具有较强的控制作用。基于以上分析，在利用复地震道计算信号的瞬时信息作为时频原子频率、相位和时延等参数的基础上，对Morlet小波时频原子的尺度参数首先进行一维寻优，在得到最佳尺度因子基础上对时频原子参数进行微调，提高了计算效率。针对模型测试了算法的有效性及在去除噪声和薄层厚度求取等方面的应用前景。

英文摘要：

Matching pursuit time-frequency analysis has better time-frequency resolution compared to short-time Fourier transform,continuous wavelet transform,and generalized S transform,but the tradi-tional greed iterative algorithm has lower computation efficiency.The Morlet wavelet is chosen as time-frequency atoms to achieve the matching pursuit due to the good property of scale parameter.The scale parameter has strong control action on the form of time-frequency atom,thus has strong control action on the matching character between signal and time-frequency atom,through comparing and analyzing the forms of time-frequency atoms based on different scale parameters and the projection values of signal onto different time-frequency atoms with the same frequency,phase and time-delay parameters but different scale parameters.The 1 D optimization for scale parameter is done with the frequency,phase and time-de-lay parameters calculated by Hilbert transform as the parameters of time-frequency atoms.Then the pa-rameters are only needed to be fine-adjusted,and the computation efficiency is improved.The effective-ness of algorithm is tested by model data,and the algorithm is also tested on application of denosing and inversion for thin layer thickness.