Jointly Verifying Ownership of an Image Using Digital Watermarking

The current image watermarking schemes only involve one key, thus can not resolve the problem of joint ownership. This paper proposes two totally new algorithms that make use of a secret sharing scheme in cryptography to address this problem. The first one applies Shamir's (2,2) threshold scheme to the watermarking algorithm. A watermark, which is a gaussian distributed random vector determined by two keys, is embedded to selected coefficients in all middle bands in the wavelet domain of an image, so that only when the two keys are put together can the ownership be verified. The second algorithm is a modification of the first one. Three random watermarks are embedded to middle bands in the wavelet domain of an image. For the watermark detection, two thresholds are set, so the watermark detector can verify partial ownership as well as full ownership. Experimental results show that both algorithms have the desired properties such as invisibility, reliable detection and robustness against a wide range of imaging processing operations.