Abstract: Considering the micromotion structure comprised of multi scatters, its micromotion parameter estimation is equivalent to the instantaneous frequency estimation of multicomponent nonstationary signal. Aiming at this problem, a novel method based on curve tracking algorithm is proposed to estimate the parameter of micromotion. First, the proposed method separates the timefrequency curves successfully in timefrequency domain with the nearest neighbor data association (NNDA) algorithm, then the timefrequency curve of each component signal is smoothed by the extended Kalman filter, and the parameter of micromotion can be estimated with the smoothed timefrequency curves. In the simulation, electromagnetic data is used to verify the validity of the proposed algorithm.