A new approach for transformation invariant image and video watermarking in the spatial domain: SSP - self spanning patterns

A variety of robust non-perceptual watermarking methods have been proposed to solve the illegal copying and proof of ownership problems. Although a number of broad claims have been made in the field of robustness of various digital watermarking methods. It is still difficult to handle combined or non-linear geometric transformations. In this paper we propose the SSP approach. We use a correlation based watermark pattern approach in the spatial domain which operates on image inherent properties when selecting the marking positions: self spanning patterns (SSP). Each watermark pattern which represents one information bit of the whole watermark suing is spanned over four image inherent properties like image edges or corners. These four points specify a polygon over which the watermark pattern is spanned using a polygon filling algorithm. After transformations we search for our points based on the image characteristics. Then we span our pattern over the points and check if there was a one or a zero embedded. Continueing that for all pattern positions we can build the whole watermark suing. The advantage of the SSPs is that transformations need not to be recognized, the marking positions and our pattern shape can be automatically found.