The invention discloses a weak characteristic information recovery system based on stochastic resonance. The system consists of a sensor, a bistable analog circuit, a recovery analog circuit and a filter module, wherein practical sensed signals of the sensor as primary data are sent to the bistable analog circuit and the like in the system. The recovery analog circuit consists of three constructed analog multiplier chips, a direct current power supply and a feedback amplifier, wherein an output end of a third multiplier is connected to the feedback amplifier to carry out feedback output for the signals. The system adopts the principle that: the signals practically sensed by the sensor are full submerged by noise and cannot be distinguished, and the signal to noise ratio of the primary weak signals input by the system is greatly enhanced through the processing such as the stochastic resonance, waveform recovery, filtration and the like. The system solves the problem of fine detection of weak signals in strong noise, breaks through the limitation that the prior low-pass filter distorts the information detection for the weak signals in the strong noise, and can realize the functions of multi-period, multi-frequency and multi-component complex weak characteristic information identification and signal reduction, and the like.