To de-noise Gaussian noise of images, an image filtering algorithm based on translation invariance wavelet transformation was proposed to overcome shortcomings of classical image filtering algorithms. The influence of wavelet decomposition function, threshold coefficients and translation space on filtering result was also discussed. The simulation results indicate that the proposed algorithm can de-noise and retain the edge of image effectively; Meanwhile, the PSNR was also improved than that of mean filtering and wavelet soft/hard threshold filtering algorithms. It's an effective Gaussain noise image filtering algorithm.