Infrared image with white Gaussian noise is processed by nonsubsampled Contourlet transform. The statistical characteristic of its coefficients is analyzed and generalized Gaussian distribution is used to describe the probability distribution for coefficients. According to characteristics of different energies in each direction of the nonsubsampled Contourlet transform bandpass subbands, a modified Bayesian threshold formula is proposed. In order to overcome the shortcoming of the soft and hard thresholding function, then a new adjustable and adaptive thresholding function is presented. Lastly, the new thresholding function is used to estimate coefficients without noise, and inverse nonsubsampled Contourlet transformation is performed to get denoised infrared image. Experimental results show that our denoising algorithm outperforms the usual wavelet threshold denoising method in peak signal-to-noise ratio and visual quality.