Brandon Migdal May 7 th 2004 Rochester Institute of Technology Advisor: Dr. Carl Salvaggio Watermark Extraction Methods For Linearly Distorted Images to.

Watermark Extraction Methods For Linearly Distorted Images to

Maximize Signal-to-Noise Ratio

Brandon MigdalBrandon MigdalMay 7May 7thth 2004 2004

Rochester Institute of TechnologyRochester Institute of TechnologyAdvisor: Dr. Carl SalvaggioAdvisor: Dr. Carl Salvaggio

OutlineOutline

Embedding ProcessEmbedding Process Research Research Extraction ProcessExtraction Process ResultsResults Future WorkFuture Work

Embedding ProcessEmbedding Process

Carrier Message

100101…

FFT

Multiplication

Inverse FFT

Scale AmplitudeZero Mean

Initial Image

FFT

+

Template DesignTemplate Design

ConsiderationsConsiderations Location DesignLocation Design ReasonsReasons

TemplateTemplate

TemplateTemplate

TemplateTemplate

Embedding ProcessEmbedding Process

Carrier Message

100101…

FFT

Multiplication

Inverse FFT

Scale AmplitudeZero Mean

Initial Image

FFT

+

CarrierCarrier

Carrier DesignCarrier Design

Random PhaseRandom Phase Generated Pseudo-Random Generated Pseudo-Random Scale based on Nyquist, and CSFScale based on Nyquist, and CSF Reasons Reasons

Embedded ImageEmbedded Image

Distortion Distortion

Research AreasResearch Areas

WindowingWindowing SubsettingSubsetting

WindowingWindowing

FFT

Windowing Technique

Substitute IntoLevel 3

Extraction Process

SubsettingSubsetting

Subset

Level 3Extraction Process

+

Extraction ProcessExtraction Process

Embedded Image

Level 3 Extraction

FFT FFT

Cross Correlation

FFT-1

Blurring Filter

SNR Extraction

Level 2 Extraction

Comparison

Image Information

Level 1 Extraction

Image Information

&SNR

Scaling Factor

X

+

Scaling Factor

X Divided by Total Number of Image Blocks

For All Image Blocks

Average Image Block

FFT

Cross Correlation

Binary Message

ResultsResults

SNR Ratios

5

10

15

20

25

30

35

0 2 4 6 8 10 12 14 16 18 20

Distortion

Sig

nal

-to

-No

ise Normal Whole

Hamming Whole

Hanning Whole

Expon. (Normal Whole)

Expon. (Hamming Whole)

Expon. (Hanning Whole)

ResultsResults

Improvement From Substeting Without Windowing

-0.00001

-0.000005

0

0.000005

0.00001

0.000015

0.00002

0 2 4 6 8 10 12 14 16 18 20

Distortion

Sig

nal

-to

-No

ise

SNR - SNR 4

SNR - SNR 16

Log. (SNR - SNR 16 )

Log. (SNR - SNR 4)

ResultsResults

Improvement From Subsetting With a Hanning Window

-1.50E-05

-1.00E-05

-5.00E-06

0.00E+00

5.00E-06

1.00E-05

1.50E-05

2.00E-05

2.50E-05

3.00E-05

3.50E-05

4.00E-05

0 2 4 6 8 10 12 14 16 18 20

Distortion

Sig

nal

-to

-No

ise

SNR - SNR 4

SNR - SNR 16

Log. (SNR - SNR 16 )

Log. (SNR - SNR 4)

ResultsResults

Improvements From Subsetting With a Hamming Window

-1.00E-05

-5.00E-06

0.00E+00

5.00E-06

1.00E-05

1.50E-05

2.00E-05

2.50E-05

3.00E-05

0 2 4 6 8 10 12 14 16 18 20

Distortion

Sig

nal

-to

-No

ise

SNR - SNR 4

SNR - SNR 16

Log. (SNR - SNR 4)

Log. (SNR - SNR 16 )

Future WorkFuture Work

WindowingWindowing SubsettingSubsetting

Reference MaterialReference Material Gonzalez, R., and R. Woods, Gonzalez, R., and R. Woods, Digital Image ProcessingDigital Image Processing, ,

Second Second Edition, 2002, New Jersey: Prentice HallEdition, 2002, New Jersey: Prentice Hall Rabbani, M., and C. Honsinger, Rabbani, M., and C. Honsinger, Data Embedding Using Data Embedding Using

Phase Phase DispersionDispersion, IEE Seminar , IEE Seminar on Secure on Secure Images and Image Images and Image Authentication Ref. No.00/039, Authentication Ref. No.00/039, 2000, Volume 5, 2000, Volume 5, pp. 1-7pp. 1-7

Xie, H. N. Hicksa, G. Kellera, H. Huangb, and V. Xie, H. N. Hicksa, G. Kellera, H. Huangb, and V. Kreinovich, Kreinovich, An IDL/ENVI implementation of the An IDL/ENVI implementation of the FFTbased algorithm for automatic image FFTbased algorithm for automatic image registrationregistration, Computers & Geosciences, 2003, , Computers & Geosciences, 2003, Volume Volume 29, pp. 1045–105529, pp. 1045–1055

Signal-to-Noise Calculations, January 2004, Signal-to-Noise Calculations, January 2004, http://emalwww.engin.umich.edu/emal/courses/SEM_lecthttp://emalwww.engin.umich.edu/emal/courses/SEM_lectureCureC W/SEM_SignalNoise.htmlW/SEM_SignalNoise.html

Questions?Questions?