Feature-based watermarks and watermark detection strategies

• US 20010036292A1
• Filed: 03/16/2001
• Published: 11/01/2001
• Est. Priority Date: 03/18/2000
• Status: Active Grant

• Current Assignee: Digimarc Corporation
• Original Assignee: Digimarc Corporation
• Inventors: Levy, Kenneth L., Decker, Stephen K.
• Granted Patent: US 7,020,303 B2
• Time in Patent Office: Days
• US Class Current: 382/100
• CPC Class Codes: G06T 1/0028 Adaptive watermarking, e.g. Human Visual System [HVS]-based watermarking
• CPC Class Codes: G06T 1/0064 Geometric transfor invariant watermarking, e.g. affine transform invariant
• CPC Class Codes: G06T 2201/0601 whereby calibration information is embedded in the watermark, e.g. a grid, a scale, a list of transformations

Abstract

Processes and apparatus for improving the state of the art for watermarking and data protection. The disclosure includes feature-based watermarks, auto- and cross-correlation techniques for determining scaling and rotation, transitions in time based watermarking, autocorrelation watermarks for images, and dynamic content scrambling of static files.

Claims

1. A feature based watermark embedding method for hiding auxiliary data in a media signal comprising:

identifying N features in the media signal, where N is an integer greater than one; and

embedding a watermark around the N features by modulating sample values in a group of samples around each feature according to a watermark signal layer.

2. The method of

claim 1

wherein the features comprise peaks of a derivative of the media signal.

3. The method of

claim 2

wherein the media signal is an image.

4. The method of

claim 1

wherein the watermark signal is a noise layer.

5. The method of

claim 1

wherein different types of watermark signals are embedded into the media signal:

a first type embedded around the N largest features in the media signal, and a second type spread around the media signal.

6. The method of

claim 5

wherein the media signal is an image.

7. The method of

claim 6

wherein the features comprises peaks of the image.

8. The method of

claim 7

wherein the features comprise peaks of the derivative of the image.

9. The method of

claim 5

wherein the second type carries a message payload.

10. The method of

claim 1

wherein the watermark signal has correlation properties that enable a watermark decoder to compensate for scaling distortion by computing auto or cross correlation of the watermarked signal and deriving scaling from positioning of peaks in a resulting signal.

11. The method of

claim 1

wherein the watermark comprises a PN sequence of symbol values of either 1 or −

1, the symbols are mapped to samples of the media signal, and transitions in phase between adjacent groups of samples corresponding to different symbols are made to change slowly.

12. A method of steganographically embedding a watermark in a media signal comprising:

forming a message comprising a PN sequence of symbols of either 1 or −

1;

mapping the symbols to samples in the media signal;

smoothly varying transitions in phase between adjacent groups of samples corresponding to different symbols.

13. The method of

claim 12

wherein the media signal is an audio signal.

14. A method of decoding a feature based watermark that hides auxiliary data in a media signal comprising:

identifying N features in the media signal, where N is an integer greater than one; and

decoding a watermark around the N features by correlating sample values in a group of samples around each feature with a watermark signal layer;

using one of the N features as a reference for decoding auxiliary data for one or more of the other features.

15. The method of

claim 14

wherein the features are peaks in the media signal.

16. The method of

claim 15

wherein the peaks are peaks of a derivative of the media signal.

17. The method of

claim 14

wherein correlation peaks generated by autocorrelation of the watermarked signal are used to compensate for scaling distortion of the watermarked signal before decoding an auxiliary data message.

18. The method of

claim 17

wherein the media signal is an image and the correlation peaks are used to compensate for rotation distortion before decoding the auxiliary data message.

19. The method of

claim 14

wherein multiple different watermark layers are decoded including watermark layers around the features and elsewhere.

20. The method of

claim 19

wherein the watermark layers employ PN sequence carrier signals detected using cross correlation or autocorrelation functions on the watermarked signal.