Abstract: According to the disadvantages of sparse coding shrinkage and Contourlet transform，a new image denoising algorithm is introduced.For that an infrared image is corrupted by additive noise with unknown variance，this algorithm is very effective.Experimental results show that this algorithm improves the SNR value，reduces the MSE value and obtains a better quality of image restoration comparing with other methods such as traditional denoising methods，sparse coding shrinkage and methods based on Contourlet transform.