Abstract: In order to ensure the security and integrity of the authenticated user’s fingerprint data transmitted via Web and increasing the credibility of the automatic fingerprint identification system (AFIS), an adaptive embedding algorithm is introduced that can hide the fingerprint minutiae data in a synthetic fingerprint image. Based on characteristics of human visual system (HVS), including texture/contract masking property and sensitivity of frequency, and combining with wavelet multi-resolution theory, a novel just noticeable difference (JND)model that can embed watermark adaptively is proposed. Experiment results show this algorithm is invisible and has good robustness.