This document discusses a method for denoising satellite images using curvelet transform and k-means clustering. Curvelet transform is used to denoise the images by taking advantage of its ability to represent lines and edges. K-means clustering is then used to segment the image into background and water regions. Bridges are then extracted based on pixel intensity differences. The methodology is tested on satellite images and the denoised images are found to have higher PSNR values compared to the noisy input images, indicating the method is effective at reducing noise from satellite images.