In this paper, a novel, yet simple, watermarking algorithm for image authentication is proposed using fractional wavelet packet transform (FRWPT) via singular value decomposition (SVD). Unlike the traditional watermarking schemes where the watermark is added to the transform coefficients, the proposed algorithm is based on embedding in the singular values (luminance) of the host image. To improve the fidelity, the perceptual quality of the watermarked image and to enhance the security of watermarking, we model an adjustable watermarking algorithm. The meaning of the word adjustable is that the watermark is embedded into the host image by taking two watermark embedding strengths, according to owner and some cryptographic conditions. Finally, a reliable watermark extraction algorithm is developed for the extraction of watermark from the distorted image. The feasibility of this method and its robustness against different kind of attacks are verified by computer simulations and comparison with the existing work. © 2011 Elsevier Ltd. All rights reserved.