This paper proposes an optimal stochastic resonance (SR) approach of the traditional quartic double-well bistable system, which is driven by a weak noisy single-frequency sinusoidal signal under low signal-to-noise ratio (SNR) circumstance. By introducing an external independent additive SR noise and optimizing the corresponding two driving parameters of the weak noisy sinusoidal signal and the external SR noise, a maximal output SNR of the SR system state variable can be reached together with SNR improvement. Simulation results verify the effectiveness of the approach. And it can certainly be applied to weak single-frequency target detection or identification problems and so on.