High noise remote sensing image denoising was a difficult problem in the field of remote sensing research. In this paper, to improve the reconstruction quality of remote sensing images, an improved iterative wavelet thresholding (IWT) algorithm is proposed on the foundation of the classical compressed sensing iterative wavelet thresholding (IWT-CS) algorithm. First, an adaptive wavelet filtering operator was proposed to remove parts of image noise, which selects wavelet coefficients of the remote sensing image in the process of image sparsity transform. Second, a descending BayesShrink threshold was put forward to select the wavelet coefficients obtained again in each iteration. Finally, an improved group sparse total variation (IGSTV) method was utilized to adjust the obtained reconstructed image to further improve the reconstruction quality of the remote sensing image. The experimental result demonstrates that the proposed algorithm was superior to the IWT-CS algorithm in terms of denoising reconstruction, which could recover a high quality remote sensing image from high noise image, and the effectiveness of the proposed algorithm was validated. In addition, the proposed algorithm could effectively protect the edges, textures, and other important feature information in remote sensing image. Under low compression sampling ratio, the proposed algorithm could still obtain relatively high peak signal to noise ratio (PSNR) and visual quality. In the satellite receive station, the proposed algorithm can be directly used to reconstruct a clear remote sensing image using a small amount of received-noise remote sensing image data.