A new technique, twice sampling stochastic resonance (SR), is proposed, and with this technique, the goal of detecting a weak signal overwhelmed in a noise is r ealized under large parameters in terms of the theory of adiabatic elimination. For the purpose of practical applications, the relative parameters are investiga ted in the detection of a weak signal based on the technique. The numerical simu lation shows that the method presented here is of potential value in the signal analysis and is expected to be applied to the practically measured data processi ng in the future.