Abstract: The method that removes the 1/ f noise in the IR image is studied. The feature of frequency and power spectral density of 1/ f noise is analyzed and the effect of wavelet transform is compared with Wiener filter. Because one of characteristics of wavelet transform is that it can remove the correlation of noise, wavelet transform is utilized to makes 1/ f noise become white noise that can be removed by selecting a reasonable wavelet base. The experimental result shows it is effective to remove 1/ f noise by applying wavelet transform and the S/N is enhanced, which is helpful to the target detection.