In the image denoising methods based on discrete wavelet transform, the generalized cross validation (GCV) algorithm has been proven to be an effective statistical way for estimating the optimal threshold and used widely to remove the image noise. However, GCV has the higher computational complexity than other denoising threshold estimating method. For the high spatial resolution remote sensing image, the GCV algorithm spends most time for computing the wavelet denoising threshold of every subband. An effective and efficient high spatial resolution remote sensing image denosing algorithm based on region of interest (ROI) and fast GCV is proposed. This new algorithm first obtains these image regions of interest (ROI) using shape adaptive integer wavelet transform (SA-IWT) and then computes the denoising threshold of ROI on the high spatial resolution remote sensing image by fast GCV algorithm. Finally, the new algorithm completes the ROI denoising using the soft-threshold merhod. The experimental results show that the new algorithm can not only first complete ROI denoising of the remote sensing image, but also reduce the computational complexity of GCV effectively. This new method is valuable for future high spatial resolution remote sensing image denoising.