Eliminating Noise of MR Image with Wavelet Transform

A new noise-eliminating processing algorithm for Magnetic Resonance (MR) image is proposed according to the different properties of image signal and noise under the multi-scale wavelet transform. This algorithm directly compares the coefficients of the adjacent wavelet domains and a filter is established for eliminating noise according to the comparison results. The computer simulation results show this algorithm has a fast operating speed and 91% noise can be eliminated. The signal-to-noise- ratio is improved from 3.1861 to 13.4784.