Abstract: ‍ ‍The precession frequency of warhead and decoy in the ballistic missile midcourse varies greatly，due to their different mass distribution. The stable and unique micro-motion characteristic of rotating targets is a current focus of the ballistic missile target recognition， which can be extracted from their micro-Doppler information. This paper takes the cone-cylindrical warhead as the research object， established a typical scattering point model and precession model of warhead， and then analyzing micro-Doppler periodicity through precession law of scattering point. This paper proposed an algorithm for the precession frequency estimating based on the time-frequency distribution of the warhead's scattering points. The algorithm applied the short-time Fourier transform to the strong scattering center unit of HRRPs and then used an adaptive threshold method to the time-frequency distribution for targets' precession frequency extraction. This paper simulated the ballistic echo data and used the data to verify the effectiveness of our algorithm. The experimental results prove that the algorithm has good anti-noise performance.