There exist some shortcomings in traditional soft and hard threshold denoising methods, such as a constant deviation between the wavelet coefficients before and after denoising, and the threshold function is not continuous etc., therefore these methods could not achieve desired denoising effect. A new threshold denoising method is presented based on the traditional hard and soft threshold methods. Denoising experiments are performed by using three threshold functions in Matlab environment. The signal to noise ratio and the size of the mean square error obtained by the three methods are compared. Simulation results show that the new threshold function not only bears good continuity, but also greatly reduces the constant deviation between the estimated wavelet coefficients and the true wavelet coefficients in the soft threshold function，which further confirms that the new threshold function can remove the white noise more flexibly and effectively.