College of Geo- Exploration Science and Technology，Jilin University，Changchun 130026，China

In process of seismic exploration，the noise of seismic signals produces serious interference． Conven- tional methods of wavelet threshold denoising cannot fully use the characteristics of seismic signals due to its limitations． There is always a certain degree of deviation between estimated value and actual value． In this stu- dy，a method of seismic data denoising is proposed，the authors use the current coefficients，the parent coeffi- cients and the neighborhood coefficients based on dual- tree complex wavelet transform (DTCWT) and related sub-band denoising model (TrivaShrink) to achieve the optimal estimation of shrinking factor and get the noise reduction of seismic records． It is found that the method is better than conventional methods of wavelet threshold denoising in removing random noise．