Abstract: As a very important part in image restoration,image denoising is to get a high quality image on the visual.Previous studies on wavelet domain wiener filtering were studied as the image has a zero mean condition,but most image is not generally as a zero mean condition in reality,so in this paper,we considered that the image doesn't has a zero mean assumption,then improved the local expected square error parameter,and created the parameter LSI.Through the comparison of the original parameter to the LSI,we carried out the threshold processing according to the small number of them,and then selected the wavelet coffieients,we made the image deniosing in the end.The simulation results showed that the improved filter had a higher peak signal to noise ratio for the image,and effectively improved the performance of the wavelet domain wiener filtering algorithm.