Abstract: A proved secure protocol for watermark verification based on perfect zero knowledge interactive proof system and bit commitments scheme is proposed. Most of the existing literature use, Cox's spread spectrum watermarking scheme or its or its modified ones. In this paper, the multiplication embedding rule is adopted, which is more robust and fits in with copyright protection. In the process of watermarking extraction, under the condition of taking into account, the different transform domains (such as DWT, DCT and DFT), communication channel performance and human visual system, generalized Gaussian distribution model and Weibull distribution model based robust optimum watermarking detectors of different transform domains are adopted. Watermarking information is committed by using bit commitments scheme, and by jointing random sequence, the watermarking embedding position information is hidden. The proposed protocol ensures that a prover soundly convinces a verifier of the presence of a watermark in certain stego-data without revealing any knowledge (including watermarking itself, embedding position, watermarking extraction keys) about the watermark which can be used by a potential attacker to modify, remove or spoil the watermarks. Moreover, various watermarking detectors in different transform domains are adopted. Thus, by using the presented simple and secure protocol, the security and efficiency of watermarks verification of the different watermarking detection schemes can be improved greatly.