Abstract: Wavelet transform de-noising only decomposes and processes the low-frequency part of the signal, neglects the decomposition and processing of the high-frequency part of the signal, which can not effectively extract the whole frequency band information of the signal, thus affecting the denoising effect of pulsar signal. In view of this, this paper proposes to use wavelet packet threshold method to analyze pulse signal, and selects different threshold functions to process the coefficients after wavelet packet decomposition. The experimental results show that the denoising effect of pulsar signal is improved by using this method, and the signal-to-noise ratio, peak signal noise and smoothness index of pulsar signal are improved correspondingly, which provides a new idea for the denoising of pulsar signal.