This paper presents an adaptive and robust invisible watermarking method for digital images. The proposed method is carried out into four phases. In the first phase, Scale Invariant Feature Transform (SIFT) is used to extract features and clustered through k-means clustering algorithm. In the second phase, around center of each cluster a minimum bounding box is formed and the underlying image contents are cropped for inserting watermark. In the third phase, each selected region is wavelet transformed. In the fourth phase, scaling parameter is calculated adaptively using middle frequency bands of each cropped image region through hyperbolic tangent function. The watermark is resized based on the size of region cropped and inserted in to the middle frequency band of all selected regions. Further, the method is more robust to RST, cropping and other attacks due to extraction of RST invariant SIFT features and insertion of single watermark in multiple locations. Experimental results prove the robustness of the proposed method against various attacks. © 2013 by CESER PUBLICATIONS.