Linear projective data embedding
AssigneeEastman Kodak Company
InventorChris W. Honsinger
Life of this patent
Estimating the term…
Abstract
In a method of embedding a hidden message in digital data, employing plurality of orthogonal random arrays of numbers as carriers for the bits of the hidden message, the random arrays are generated by applying a plurality of unique cyclical shifts to a single random array. In a preferred embodiment, the method is applied to digital image data and the cyclical shifts are defined as locations in a shift template, wherein the shift template has a highly peaked autocorrelation. The shift template is employed to recover the hidden message from the image without the need for calibration signals to indicate the boundaries of the hidden message.
Claims5
In a method of embedding a hidden message in digital image data, wherein the method is described by:
- I ′ ( x , y ) = I ( x , y ) + ∑ i = 1 n b i · R i ,
- where I′(x,y) is the digital data containing the hidden message,
- I(x,y) is the digital image data, n is the number of bits, where n is equal to or greater than 2,
- bi are the bits of the hidden message, and
- Rj are random arrays of numbers having the property that R i · R j = { 1 where i = j 0 where i ≠ j ,
- the improvement comprising:
- Ri=C(x−αi,y−βi)
- where C(x,y) is a array of random numbers, and
- (αi,βi) are offsets having the property that (αi,βi)≠(αj,βj) for i≠j;
- wherein the distance ((αi−αj)2+(βi−βj)2)0.5 between any two offsets is greater than the autocorrelation width of C(x,y).
A computer-readable medium having a program embodied thereon, said program carrying out a method of embedding a hidden message in digital image data, wherein the method is described by:
- I ′ ( x , y ) = I ( x , y ) + ∑ i = 1 n b i · R i ,
- where I′(x,y) is the digital data containing the hidden message,
- I(x,y) is the digital image data, n is the number of bits, where n is equal to or greater than 2,
- bi are the bits of the hidden message, and
- Ri are random arrays of numbers having the property that R i · R j = { 1 where i = j 0 where i ≠ j ,
- the improvement comprising:
- Ri=C(x−αi,y−βi),
- where C(x,y) is a array of random numbers, and
- (αi,βi) are offsets having the property that (αi,βi)≠(αj,βj) for i≠j;
- wherein the distance ((αi−αj)2+(βi−βj)2)0.5 between any two offsets is greater than the autocorrelation width of C(x,y).
Dependencies are inferred from the claim text.