Steganography on point-sampled geometry

We present two new schemes for digital steganography of point-sampled geometry in the spatial domain. To the best of our knowledge, our schemes are the first in the literature for such types of geometry. Both schemes employ a principal component analysis to translate the points’ coordinates to the new coordinate system, resulting in a blind approach. In the first scheme, we establish a list of intervals for each axis according to the secret key. We then embed a bit into each interval by changing the points’ position. In the second scheme, we locate a list of macro embedding primitives (MEPs), and then embed c   bits (2⩽c⩽6)(2⩽c⩽6) at each MEP, instead of a single bit as in the first approach. We validate our schemes with various models in terms of capacity, complexity, visibility, and security. Experimental results demonstrate that the proposed schemes are robust against translation, rotation, and scaling operations. In addition, our schemes are fast and can achieve high data capacity with insignificant visual distortion in the stego models.