Most watermarking methods presented so far belong to the category of symmetric watermarking in that the secret key is undesirably revealed during the watermark detection process. In view of this security leakage, zero-knowledge watermark detection (ZKWD) has been introduced without obviously revealing the secret information. However, the existing ZKWD protocols still suffer from some challenging problems, which will be addressed in this paper. First, a zero-knowledge watermark detection protocol is presented based on our robust image watermark scheme so that robustness against removal and geometric attacks can still be retained. The aim is to extend the capability of our own system to satisfy more watermarking requirements as possible. Second, the watermarks revealed for zero-knowledge detection are. still secured by a secret key-based shuffling function so that the verifier cannot have the knowledge about the hidden watermarks. Third, we show how our protocol can be operated in a non-interactive way. Finally, in order to prevent from cheating prover, the Hamiltonian cycle is introduced in a media data before its publication. Thus, the overall characteristics distinguish our protocol from the existing protocols significantly.