In order to improve the counting accuracy of laser particle counter, a signal preprocessing method based on matched filter is designed. The characteristics of the particle signal are analyzed, and the typical waveform of the particle is obtained by the method of normalization and accumulation. Using the typical signal as a matching template, the original signal of laser particle counter and the matching template are discretely convolved to obtain an output signal with a high signal-to-noise ratio. The experimental results show that the signal-to-noise improvement ratio is better than 1.6 for particle signals with original amplitude of 55~262 mV and root mean square value of noise of about 6.74 mV. This method effectively improves the signal-to-noise ratio of the output signal, and provides a good foundation for the accurate detection and recognition of particles.