The performance of various estimators, such as maximum a posteriori (MAP), strongly depends on correctness of the proposed model for distribution of noise-free data. Therefore, the selection of a proper model for the distribution of wavelet coefficients is very important in wavelet based image denoising. This paper presents a new image denoising algorithm based on the modeling of wavelet coefficients in each subband with a mixture of Laplace random variables. Indeed, we design a MAP estimator which relies on mixture distributions. Using this relatively new statistical model we are better able to capture the heavy-tailed nature of wavelet coefficients. The simulation results show that our proposed technique achieves better performance than several published methods, both visually and in terms of root mean squared error (RMSE).