The properties of the output signal of heterodyne laser Doppler vibrometer (HLDV) under random noise disturbance are studied by means of wavelet transform. Considering the principle of HLDV and the influence of white noise in this system, the modulus maximum and the coefficient correlation algorithms are analyzed and improved. Also, based on the theory of the threshold enhancement, a new threshold function is proposed and used to analyze the effects of multi-scale method on the reconstruction of the oriental signal and improvement of signal-to-noise ratio (SNR). The results show that the improved modulus maximum, coefficient correlation and threshold enhancement algorithms present much more excellent performances in extracting the vibration parameters of the micro-vibration targets.