A hybrid Fourier-wavelet image de-noising using Local Gaussian Scale Mixtures Model is proposed.The de-noising method has overcome the disadvantage of Fourier transform and wavelet transform.It takes into account the correlation between wavelet coefficients of noise.The Local Gaussian Scale Mixtures Model is used to estimate the wavelet transform coefficients.Experiment results reveals that the proposed method can perform a good performance on de-noising.As the de-noise result of traditional wavelet is related to the wavelet basis function,this method overcome the limitation.Through experiments with images contaminated by additive random noise,we demonstrate that the performance of this method substantially surpasses the previously published methods,both visually and in terms of signal to noise ratio.