Abstract: Based on exploiting the correlations among the image wavelet decomposition coefficients in a subband, an adaptive statistical model for wavelet image coefficients was presented and applied to the image denoising of Remote Sensing Image． Each wavelet coefficient was firstly modeled as a random variable of a generalized Gaussian distribution (GGD)，then，based on the algorithm of the wavelet soft threshold denoising and according to the characteristics of spatial clustering of wavelet decomposition coefficients, a new local adaptive algorithm was proposed and applied to restore the noisy images by estimating the coefficients with maximum a posteriori probability rule (MAP).The algorithm was applied to denoise the noisy Remote Sensing Image of Maoergai area in the upper Minjiang where contains typical vegetation and soil．Simulation results showed that the higher peaksignal to noise ratio and the better visual effects were obtained as compared to other image denoising methods．