Abstract: To study the security protection of remote sensing images, a robust watermarking algorithm based on SURF feature regions was proposed. Initially, feature points were extracted from original image by using SURF detector, and points with stronger robustness were further selected to construct non-overlapping feature regions and image normalization was performed. Then integer wavelet transform was applied on the normalized regions, and corresponding position coefficients were selected according to the generated template. Finally, the watermark was embedded repeatedly into every feature region using quantization method. Watermark detection could be done directly from the attacked image without correction, and also realize the blind extraction. An edge pixels replacement scheme was used to ensure the data accuracy, so the watermark had no effect on the application of remote sensing images. Experimental results show that the proposed algorithm is not only invisible, but also robust against common image processing operations and geometric attacks, so it can protect remote sensing images security effectively.