Laser Doppler Velocimetry (LDV) has been widely used for many years in fluid mechanics to measure particle velocity. However, the signal of LDV is weaker in most applications. One challenging issue is how to effectively discover the weak characteristics from the signal submerged in noise. To address the issue, a method based on Stochastic Resonance is proposed in this paper. Using the SR method, the noise may be weakened and weak characteristics highlighted, and therefore velocity can be is obtained. FFT and the method are adopted to process simulated laser Doppler signals in circumstance of MATLAB. The input and the output signal-to-noise ratios of the bistable system demonstrate the effectiveness of the SR method.