In order to solve the problem of processing non-stationary signal of blasting vibration, the signal is analyzed by wavelet transform and Hilbert-Huang transform. This paper presents reconstructed vibration signal based on wavelet base function of sym5 and Empirical Mode Decomposition is very consistent similar to original vibration signal with relative error are 10-12 and 10-15 respectively. By comparing relative error and curve features between raw signal and reconstructed signal of blasting seismic wave, the conclusion that wavelet transform and Hilbert-Huang transform both have high performance to detect the location of catastrophe point and reflect the time-domain Characteristics of signal is obtained. The distribution characteristics about energy of blasting vibration signal with frequency and time is investigated, which show that energy density spectrum on Hilbert-Huang transform is in accordance with wavelet spectrum on sym5 wavelet basis function where the energy mainly concentrates on the range of 10~80Hz and 0.125~0.625s.