Stochastic resonance(SR) enhances the nonlinear system behavior with the assistance of noise,including the sensitivity and selectivity of the response to the exterior stimulus. The energy-transfer mechanism makes the weak information revealed in the output spectrum,while the time-waveform is distorted. The distortion analysis was made both from the particle's dynamics and signal processing. The factors causing the deviation in the output are presented and the function of the recovery system is proposed. By the investigation of the particle's motion track in the bistable system and the suggested recovery system,the influences of noise and system parameters on the recovery course were discussed. Moreover,the pulse distortion appearing the recovery waveform caused by the particle's transitions at the bistable potential' inflexions was explained. Due to different characteristics,cascaded-bistable SR or mono-stable SR was introduced to process different types of signals. The final recovery signal is just the suggested recovery system's response to the SR output. Meanwhile,the recovery system is optional,as parameter-tuned or parameter-fixed. Since the method requires no average processing,it is applicable to a single sample with limited length. The numerical simulations reveal that the SR recovery method can recover the waveform containing weak information submerged in noise effectively. The engineering application to the vibration analysis of metal cutting chose the combination of mono-stable SR and the parameter-fixed recovery system. Because the optimal SR state is not required strongly,the system parameters are tuned in a wider range than the traditional SR processing methods. Stochastic resonance (SR) enhances the nonlinear system behavior with the assistance of noise, including the sensitivity and selectivity of the response to the exterior stimulus. The energy-transfer mechanism makes the weak information revealed in the output spectrum, while the time-waveform is distorted. The distortion analysis was made from the particle's dynamics and signal processing. The factors causing the deviation in the output are presented and the function of the recovery system is proposed. By the investigation of the particle's motion track in the bistable system and the suggested recovery system, the influences of noise and system parameters on the recovery course were discussed. Moreover, the pulse distortion appearing the recovery waveform caused by the particle's transitions at the bistable potential 'inflexions was explained. Due to different characteristics, cascaded-bistable SR or mono-stable SR was introduced to different types of signals. The final recovery signal is just the suggested recovery system's response to the SR output. Meanwhile, the recovery system is optional, as parameter-tuned or parameter-fixed. Since the method requires no average processing, it is applicable to a single sample with limited length. numerical simulations reveal that the SR recovery method can recover the waveform containing weak information submerged in noise effectively. The engineering application to the vibration analysis of metal cutting chose the combination of mono-stable SR and the parameter-fixed recovery system. state is not required strongly, the system parameters are tuned in a wider range than the traditional SR processing methods.