本文提出了一种多分量线性调频信号的参数估计方法。基于过完备Gabor字典的Matching Pursuit算法，可以将信号表示为Gabor原子的线性组合。这些原子有效的揭示了信号的内在时频结构特征，是信号的一种稀疏表示。本文直接利用分解得到的稀疏信息对信号中调频分量的调频率、初始频率和结束频率进行估计。仿真结果显示，该方法适用于存在强有意干扰或者有色噪声的环境。

In this paper, a novel algorithm for parameter estimation of multicomponent chirp signal in complicate noise environment is proposed. By the matching pursuit algorithm, the signal is first decomposed into Gabor atoms which provide sparse information that represent the signal time-frequency signature. Hough transform is then directly used to estimate chirp rate, start frequency and stop frequency of chirp components without computing the time-frequency distribution. Simulation results show that this algorithm is capable of estimating parameters of multicomponent chirp signals even in the presence of powerful intended interference and colored noise. Comparing to other techniques, the algorithm is computational efficient and makes it possible to apply in real-life problems.