Method for reversible image watermarking based on integer-to-integer wavelet transform

1. A method for embedding watermarks based on integer-to-integer wavelet transform, the method comprising the steps of:

(A) dividing an original image (X×Y) to a plurality of image blocks (M×N);

(B) selecting image blocks for embedding a location information that indicates image blocks to be watermarked;

(C) embedding the location information into the image blocks selected in the step (B);

(C-1) creating a matrix (P) whose elements indicate the number of bits that can be shifted in each image block; and

(D) embedding watermark and side information into remaining image blocks which are not selected in the step (B),

wherein the side information includes the matrix (P) and a set of original LSBs (Bo) replaced during the embedding of the location information.

2. The method as claimed in claim 1 ,

wherein the location information is a location matrix (L) whose size [(X/M)×(Y/N)] is same as the size of the number of image blocks,

wherein the location matrix (L) is constructed by,

L

⁡

(

i

,

j

)

=

{

i

,

⁢

if

⁢

⁢

P

⁡

(

i

,

j

)

>

0

0

,

⁢

otherwise

where 0≦i≦(X/M) and 0≦j≦(Y/N).

3. The method as claimed in claim 1 ,

wherein the size of the matrix (P) is same as the size of the number of image blocks [(X/M)×(Y/N)],

and wherein the step (C-1) further includes the steps of:

(a) Setting an induced capacity (R) and all elements of the matrix (P) to zero;

(b) Searching for the bit-shiftable blocks (S ij ) among the image blocks;

(c) applying 1-bit shifting to all the bit-shiftable blocks (S ij ) searched in the step (b);

(d) Searching for the block (S îĵ ) in which the watermark embedding may lead to the lowest mean-squared error among the blocks obtained by the step (c);

(e) letting the induced embedding capacity (R) incremented by the number of shiftable blocks and letting the value of P(îĵ) incremented by 1;

(f) determining whether the embedding capacity (R) is equal to or greater than the number of bits of watermarks and side information to be embedded;

(g) performing the steps through (c) to (f) in case that the embedding capacity (R) is smaller than the number of bits of watermarks and side information to be embedded.

4. The method as claimed in claim 3 ,

after the step (f), further comprising the step of

(f′) deleting the block (S îĵ ) from the blocks (S ij ) in case that the value of P(îĵ) is equal to p max or that the block (S îĵ ) is not bit-shiftable.

5. The method as claimed in claim 3 ,

wherein the step (f) is performed by the following inequality:

R≧|B m |+|B o |=|B m |+B l |=|B m |+( XY )/( MN )

where B m is a set of message bits, B l is a set of bits in the location map (L), B o is a set of original LSBs replaced during the embedding of the location map via LSB-substitution, and | | is the cardinality of a set.

6. The method as claimed in claim 1 ,

wherein the step (B) is performed by the following inequality:

s min −E WN ( m,n )≦ S M ( m,n )≦ s max −E WP ( m,n )

for 0<m<M and 0<n<N,

where s min is a minimum pixel value, s max is a maximum pixel value, S is the image block, S M is a modified pixel block,

and wherein E Wp and E WN are given by the following equations:

E

WP

=

∑

i

,

j

∈

(

HL

1

⋃

LH

1

⋃

HH

1

)

⁢

1

2

⁢

{

Q

ij

+

ABS

⁡

(

Q

ij

)

}

E

WN

=

∑

i

,

j

∈

(

HL

1

⋃

LH

1

⋃

HH

1

)

⁢

1

2

⁢

{

Q

ij

-

ABS

⁡

(

Q

ij

)

}

where HL1, LH1 and HH1 are an high-low, low-high and high-high wavelet sub-band respectively, Q ij =fDWT2 −1 (0 ij ) and the matrix 0 ij has only one non-zero element of value 1 in the ith row and the jth column.