In this study, an image denoising algorithm is proposed by combining Tetrolet transform and active random field (ARF). An image with Gaussian noise is decomposed with Haar wavelets, and the ARF algorithm is used to reduce Gaussian noise in the wavelet domain. After inverse wavelet transform, the Tetrolet transform is used for further denoising. The proposed method is compared with other denoising algorithms including wavelet algorithm, Tetrolet algorithm, Markov random field based algorithm and ARF based algorithm. Experimental results indicate that the proposed approach can effectively reduce Gaussian noise at various levels, and achieve better results than other algorithms.