A methods using stationary wavelet transform based on multi-scales total least squares for image denoising is proposed.The noisy image is decomposited using stationary wavelet.For the high frequency components of image decomposition,the wavelet coefficients are estimated using total least squares.The interscale correlations of wavelet coefficients are considered.The multi-scales total least squares for estimating wavelet coefficients is proposed.Then the high frequency coefficients of signal are obtained.The denoised image is obtained through inverse wavelet transform.The experimental results show that the image can be denoised effectively based on total least squares using the interscale correlations of wavelet coefficients.The SNR and the visual quality are improved substantially.