In order to solve the problem that rolling bearing fault features were difficult to be extracted under strong noise background， parameter optimized variational mode decomposition （VMD） and maximum correlation kurtosis deconvolution （MCKD） were proposed to extract rolling bearing fault features. Firstly， the original signal was decomposed by the optimal combination of parameters obtained by offline optimization of the VMD parameters using the improved sparrow algorithm. Secondly， in order to screen and reconstruct each IMF after decomposition， a new screening metric was constructed based on the envelope spectrum peak factor and sample entropy. Then， the reconstructed signal was augmented with MCKD optimized by the online method of the improved sparrow algorithm. Finally， the bearing failure frequency information was extracted from the enhanced signal by envelope demodulation analysis. Simulation and experimental results show that the proposed method is able to enhance the shock components submerged in the strong noise and effectively extract rolling bearing fault features.