DCT-Domain Watermarking

Chiou-Ting Hsu and Ja-Ling Wu, "Hidden digital watermarks in image

s," IEEE Trans. On Image Processing, vol. 8, No. 1, January 1999

Multimedia Security

2

A generic image watermarking system

• Requirements– Invisibility– Robustness – No ambiguity

Watermark

EmbeddingDistortions•Image processing•Compression•…...

ExtractionWatermark

Comparison

3

Advantages of this scheme

• Advantages:– Semantically meaningful watermark pattern– Good perceptual invisibility– Acceptable robustness– Various user-selected options– Reasonable complexity/execution time

4

System overview

• Watermarks are randomly permuted to spread their spatial relationship, and then embedded in the DCT domain of the host image, with consideration of invisibility/robustness

5

Block Diagrams of the Original Algorithm

Original Image

FDCT

Embedding(Polarity Reversing)

IDCT

Watermark Image

Pseudo-random Permutation

Block-based Mapping

Watermarked Image

6

Block Diagrams of the Original Algorithm

Original Image

FDCT

Suspected Image

FDCT

Extract the Permutated Data (XOR)

Reverse Block-based Permutation

Reverse Pseudo-random Permutation

Extracted Watermark

7

Block DCT/IDCT

• Advantages– Fast – Suitable for robustness against JPEG

compression

• Disadvantages– Block effect– Effect of picture cropping

8

Semantic Meaningful Watermarks

• Watermarks can be verified with naked eyes by understanding the semantics of the extracted watermark patterns

The seal of CML (in Chinese characters)

9

Pseudo-Random Permutation

• A n-bit Linear Feedback Shift Register (LFSR) is used to generate the maximal length (2n-1) sequence

+

The 14-bit Shift Register thatpermutes 1-16384

Original Watermark

Permuted Watermark

10

Block-based Mapping

• Watermark blocks with more signal pixels are embedded into image blocks with higher variances– to achieve better perceptual invisibility.

11

Embedding(1/2)

• Choices of embedding positions within each block:

– Low-frequency• Bad invisibility

– High-frequency• Bad robustness

=> Middle-frequency• Fix positions in each block

0 1 5 6 14 15 27 282 4 7 13 16 26 29 423 8 12 17 25 30 41 439 11 18 24 31 40 44 5310 19 23 32 39 45 52 5420 22 33 38 46 51 55 6021 34 37 47 50 56 59 6135 36 48 49 57 58 62 63

Embedding 4 watermark pixels

12

Embedding(2/2)

• Polarity: the inequality relations between the scaled DC value and the selected AC coefficients

otherwise

QQrScaleFacto

DCjiQ

jiQ

jiACif

jiP

,0

)0,0(*)0,0(*

),(*),(

),(,1

),(

• Effects of the JPEG quantization table are also considered

13

Polarity Reversing

• Polarity: the inequality relationship between DC & corresponding AC values within each DCT block.

0 1 0 1 0 0 1

1 1 0 0 1 0 1

watermark bits:

polarity:

reversed polarity: 1 0 0 1 1 0 0XOR

Embedding Extraction

14

Extraction

• Exclusive-or (XOR) operations are performed on the two polarity patterns to obtain the permuted binary data.