In this paper, as an extension of the wavelet transform, a nonlinear multiwavelet transform is introduced. The advantage of this approach is that the high frequency components of the filtered signal can be enhanced properly by adjusting a parameter that is used to control the nonlinear properties of the multiwavelet transform. Also, an optimal soft thresholding is developed. Example shows that based on the nonlinear multiwavelet transform, the optimal soft thresholding is better than the soft thresholding in smoothing noise.