X. Qi and J. Qi (USA)
Digital watermarking, feature-point-based restoration, human vision system, and rotation, scaling, translation attacks.
This paper presents a feature-point-based rotation, scaling, and translation (RST) resistant digital watermarking scheme. The improved Harris detector locates the robust and important feature points, which are further used by the acute triangle matching to determine the possible geometric attacks for reducing synchronization errors. The human visual system compatible mid-frequency embedding strength is also adaptively determined using the modified modulation transfer function. The blind spread-spectrum-based error correcting coded watermark embedding and retrieval scheme is finally applied to each 8×8 discrete cosine transform (DCT) subblock of the host image. Experimental results demonstrate the proposed system is more robust against geometric and common image processing attacks than other feature-point-based and template-based approaches.
Important Links:
Go Back
