Adaptive prediction filtering for digital watermarking

• US 20030177359A1
• Filed: 01/21/2003
• Published: 09/18/2003
• Est. Priority Date: 01/22/2002
• Status: Active Grant

• Current Assignee: Digimarc Corporation
• Original Assignee: Digimarc Corporation
• Inventors: Bradley, Brett A.
• Granted Patent: US 7,231,061 B2
• Time in Patent Office: Days
• US Class Current: 713/172
• CPC Class Codes: G06T 1/0064 Geometric transfor invariant watermarking, e.g. affine transform invariant
• CPC Class Codes: G06T 2201/0051 Embedding of the watermark in the spatial domain
• CPC Class Codes: G06T 2201/0061 Embedding of the watermark in each block of the image, e.g. segmented watermarking
• CPC Class Codes: G06T 2201/0065 Extraction of an embedded watermark; Reliable detection
• CPC Class Codes: H04N 1/32149 Methods relating to embedding, encoding, decoding, detection or retrieval operations
• CPC Class Codes: H04N 1/32331 Fragile embedding or watermarking

Abstract

The disclosure describes methods for adaptive prediction filtering for digital watermarking applications. The predictive filter is designed to predict the digital watermark signal. In one implementation, there are two stages of prediction filtering. A first stage operates on a suspect signal before detecting any watermark in that signal. This stage adapts the predictive filter to local signal characteristics. A second stage uses known watermark signal elements to adapt a non-linear filter used in extracting message symbols from the suspect signal. This stage produces detection statistics based on a comparison of expected and measured message symbol values, and uses these statistics to adapt the parameters of a predictive filter. This adapted prediction filter is then applied to extract message symbol estimates.

Claims

1. A method of digital watermark detection comprising:

computing signal characteristics of a host digitally watermarked signal;

adapting parameters of digital watermark prediction in the watermarked signal based on the signal characteristics.

2. The method of claim 1 wherein the digital watermark prediction predicts an estimate of an embedded digital watermark signal element based on comparison of a host signal value and one or more other related values.

3. The method of claim 1 wherein the parameters correspond to a type of prediction filter that is selected based on the signal characteristics.

4. The method of claim 1 wherein the digital watermark prediction predicts an estimate of an embedded digital watermark signal element based on comparison of a host signal value and one or more other related values, and the parameters correspond to a type of prediction filter that is selected based on the signal characteristics.

5. A tangible medium on which is stored instructions for performing the method of claim 1.

6. A method of digital watermark detection comprising:

computing observed values of a variable, unknown watermark signal in a host digitally watermarked signal;

comparing the observed values of the variable, unknown watermark signal with known watermark signal values at corresponding portions of the watermarked signal;

based on comparing the observed with the known values, adapting digital watermark prediction used to extract a message from the unknown watermark signal such that the prediction is weighted according to an extent of the agreement between the observed and known values.

7. The method of claim 6 wherein the digital watermark prediction predicts an estimate of an unknown watermark signal element based on comparison of a host signal value and one or more other neighboring values.

8. The method of claim 6 wherein the known watermark represents a previously detected watermark layer in the watermarked signal, and the unknown watermark represents a subsequent watermark layer to be detected in the watermarked signal.

9. The method of claim 8 wherein the previously detected watermark layer comprises a watermark variable message payload that has been error corrected to compute the known watermark signal.

10. The method of claim 8 wherein the previously detected watermark layer comprises a synchronization watermark.

11. The method of claim 8 wherein the subsequent watermark layer carries a message payload that varies from one block to another in the watermark signal to provide a block identifier per block.

12. The method of claim 6 wherein the method is used iteratively to detect a digital watermark, and with each iteration, data on reliability of portions of the digital watermark is used to weight the prediction in a subsequent iteration such that more reliable portions are weighted more heavily than less reliable portions.

13. The method of claim 6 wherein detection statistics based on a comparison of observed relative to known watermark signal values are used as a metric for detecting alteration of a watermarked signal.

14. The method of claim 13 wherein the method is used to distinguish a copy from an original printed document from a digital watermark embedded in the document.

15. A tangible medium on which is stored instructions for performing the method of claim 6.

16. A method of digital watermark embedding in a digital watermarking system where a digital watermark is extracted by predicting a digital watermark based on a comparison of values of signal elements in a digitally watermarked signal, the method comprising:

computing an initial value of a digitally watermarked signal element in a host signal;

adjusting a difference between the initial value of the digitally watermarked signal element and other elements in the digitally watermarked signal used to extract the digital watermark such that an embedded signal watermark is optimized for extraction by prediction based on desired embedding constraints; and

repeating the computing and adjusting to optimize embedding of additional data in the watermarked signal.

17. The method of claim 16 wherein the adjusting is performed for selected parts of the digitally watermarked signal to enable the digital watermark to be optimized for detection of alteration.

18. The method of claim 17 wherein the digital watermark is embedded such that watermark detection using prediction is optimized to distinguish between an original and a copy of a watermarked document.

19. The method of claim 16 wherein the digital watermark is optimized to maximize digital watermark signal detection statistics for a desired amount of perceptual distortion.

20. A tangible medium on which is stored instructions for performing the method of claim 16.